# Changelog Source: https://docs.samplevault.ai/changelog Product updates and announcements [//]: # "" [//]: # " ## v2.23.0" [//]: # [//]: # " ### Bug Fixes & Performance" [//]: # " - Fixed clicks when switching samples or seeking through a playing waveform, especially on sustained bass sounds" [//]: # " - Fixed Boundary declick softening sample attacks while disabled and missing natural endings while enabled" [//]: # " - Fixed creating a new layout failing for some existing layouts" [//]: # [//]: # "" ## v2.22.1 ### Bug Fixes & Performance * Fixed showing a sample in the File Browser opening the panel without scrolling to and focusing the sample on the first try * Fixed musical keys displaying without a space before major or minor * Fixed the filter sidebar closing when dragged below its minimum width ## A new search engine ### Highlights * **Search understands more**: "double bass" finds samples that sound like one even when the filename doesn't say so, "drum loop no kick" leaves the kick out, "C minor chord" puts samples in that key first, "glitch drums" returns drums rather than glitch effects, and pack or creator names work in folders that haven't been deep tagged * **Ready-made layouts**: the plus button in the sidebar opens arrangements for Browse, Produce, Explore, Inspect, Visualize, Reference, Sketch and Organize, already filled with the panels each job needs, or bare grids under Customize. Panels move between slots without carving out new ones, and each layout in the sidebar shows a small map of its arrangement * **Find similar sounds in the search overlay**: use any result as a reference, chain from one similar sound to the next, and compare overall, timbre, or frequency-band matches without returning to the main window * **Drag a region out**: select part of a sample in the Waveform, Spectrogram, or X-Ray panel and drag it straight onto a DAW track * **Quicker first-run setup**: the questions fit on one screen and can be skipped, and setup finds the folders your sample services already use (Splice, Loopcloud, ADSR, Noiiz) so you tick what to add instead of browsing for each path * **Sign in without leaving the app**: type your email, get a six-digit code, type it back in, with Google, Discord and the browser flow still one click away * **Smoother on large libraries**: no more freezes when Similar Samples refreshes, when a long sample plays with the Waveform or Spectrogram open, or shortly after launch - **Picking a panel reads like a shelf**: a new tab lays its panels out as cards once there's room, and stays a compact list when there isn't - **Prompt ideas on demand**: open suggested searches from the Library search bar, reroll for another set, or dismiss them without changing your search - **Your key needs attention**: when your own OpenRouter key runs out of credit, is revoked, or is blocked by its privacy settings, a card names the cause and links to your API key settings * Fixed samples at 96 kHz and above reading as having no sub-bass, which kept them out of low-end filters and band matches * Fixed missing cloud sync credentials blocking the app at startup * Fixed a full disk leaving you with no explanation when saving or searching stopped working * Fixed the Sample Map staying on "Optimising sample map" indefinitely * Fixed the Tape waveform not filling the view, and changing the Tape buffer length discarding audio already captured * Fixed files longer than ten minutes leaving library status stuck short of complete; they now classify from their opening seconds, though tempo and key stay empty for them * Fixed the Reprocess button in a stalled processing status not queueing the missing work * Fixed background organization, the Map, and search indexing failing when samples were deleted mid-run * Fixed stepping past the last sample not scrolling the list back to the top * Fixed the sign-in screen waiting forever when a browser sign-in was never finished * Fixed the browser demo showing a Professional Suite prompt instead of opening the demo library * Fixed assistant searches narrowing broad requests with invented key, tempo, and shortlist values ## Search over your DAW Hide the main window and Sample Vault doesn't go with it. Quick Actions has become a slim search bar that stays on top of your DAW: type, filter, deep search, grab audio, drag the keeper onto a track. ### The search overlay The always-on-top overlay is a search bar you type into directly, rather than a pill that opens a separate palette. Find a sample and drag it into your session, or go deeper without switching windows. * Results audition as you arrow through them, and you drag the one you want straight onto a track * Narrow with the same filters and saved presets as the library * Send a query as a Deep Search or a question, and keep refining in place with the picks landing in the results below * **Tape and sync built in**: grab the last few seconds of system audio, and reach the same tempo, key and Ableton Link controls as the main app * **Keyboard only**: arrow keys step from the search field onto the results, Space replays the one you land on, Esc puts you back in the field, and Enter copies the highlighted sample and dismisses the overlay so you can paste it into your DAW * Appears whenever Sample Vault isn't the focused window by default, with always-on or off in **Settings** and a toggle in the tray * Shortcuts for filters, Deep Search, Tape, settings, grab and recalling the main window, listed under [Keyboard Shortcuts](/reference/keyboard-shortcuts) * New installs start Sample Vault with your computer by default, so the overlay is there when you open your DAW, and existing installs are left alone Deep Search and the assistant in the overlay need a plan that includes them. Instant search and grabbing audio into Tape work even when you're signed out. ### User Experience * **Library status now reports what your library can do** rather than which internal stage it sits on, a row per capability with how far it reaches, and the same readout per folder in **Settings → Sample Directories** * **Deep tagging can be stopped and resumed** from **Library status** or **Settings → Organization**, and reports which folder it's on and roughly how long is left instead of counting internal steps * **Activity Log is now the Processing log**: it opens on failures, names each background job the same way everywhere, and points at Library status for overall progress * **Read the whole error**: a failure in the Processing log expands to its full text, which you can select or copy in one click * **Layout shortcuts**: Alt+1 through Alt+9 (⌥ on macOS) switch between your layouts, top to bottom in the sidebar * **Check connection**: a new button in **Settings → API Keys** tests your own OpenRouter key against every model Sample Vault runs on it and names what's blocking each one * **Report a problem from the startup screen**: when the app can't finish starting, one button sends us the error and your recent logs * **Deep Search stays on the search bar**: the button sits beside the search field at all times, disabled until you describe a sound * **Keyboard activation**: project cards on the Sessions board and the selected region on the Tape timeline respond to Enter and Space once focused * **Middle-drag to pan**: holding the middle mouse button drags the Map, Circle and Scatter views around from anywhere, including from on top of a sample * **Cloud backup follows your plan**: it stops when you move to a plan that doesn't include it, and reconnects to the same cloud library if you move back * Cloud sync on Professional plan is now opt-in * Storage-full, out-of-allowance and paid-feature prompts read the same wherever they appear * Moving duplicates to the Trash shows a running count and can be cancelled, and names the files the system refused at the end ### Bug Fixes & Performance * Fixed saving a Tape grab, an X-Ray or waveform region, or a Sketch render into your library failing with a save error * Fixed deep tagging a single folder on Windows failing the moment it started * Fixed deep tagging re-running folders it had already tagged, at full cost * Fixed a deep-tag run repeating the same batch forever when a request came back empty * Fixed a blocked or unfunded OpenRouter key reading as a slow deep-tag run rather than a broken one * Fixed **Move files to Trash** on the Duplicates screen abandoning the whole sweep when the system refused a single file * Fixed the library's add-filter menu closing the category, tempo and sound-profile pickers the instant they opened * Fixed the volume slider only affecting sample preview, leaving the overlay, Tape and Sketch at full level * Fixed upgrading an existing subscription to Professional Suite never setting up cloud backup or browser access * Fixed the title-bar status icon going idle while classification and search indexing were still running * Fixed the processing time estimate reading roughly three times too long, and still quoting one after processing had stalled * Fixed **Retry failed** requeuing the whole queue instead of the jobs the confirmation counted * Fixed failed upload and re-index rows reading "Unknown sample" instead of the file they belonged to * Fixed database backups being deleted instead of kept when they failed their integrity check * Fixed the Tape button in the title bar needing a second click to open the window * Fixed typing a key and scale together in the search bar, like `@CMaj`, dropping the scale * Fixed sample previews on macOS becoming increasingly delayed or falling silent during long browsing sessions * Moving duplicates to the Trash on macOS now takes seconds instead of up to an hour, and no longer asks for permission to control Finder * Opening a directory in Sample Directories shows its first page in about half the time on a large library Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.20.6 ### Bug Fixes & Performance * Fixed the "Updating library" step after an update hanging for hours on very large libraries that had already been deep tagged ## Split a track into stems X-Ray can now pull a reference track apart into vocals, drums, bass and everything else, then match each part against your library on its own, so a drum lane finds your drums instead of everything at once. Separation runs on your machine and the track never leaves it. * Six lanes per track: vocals, drums, bass and other, plus groove (drums and bass together) and the full instrumental * Drag a region on any stem lane to find the sounds you already own that match that part * Splitting a four-minute track takes a few minutes and deliberately leaves most of your CPU free Stem splitting is part of the paid plans, and the first split downloads a one-time 300 MB file that stays on your machine afterwards. Dropping a track and searching against the whole mix is unchanged and still free. ### Your DAW projects, in the browser Projects now have a place to search them, the sounds they used, and a way back into them from any sample. * **Import from the title bar**: the Projects dropdown imports your DAW projects from a folder without a trip to Settings * **Find and sort**: search the Projects board by name, genre, key or mood, and order it by recently modified, recently created, or name * **Samples tab**: every sound from a project's sessions as a playable row, marking which are in your library and which aren't * **Sessions tab**: when a sample turns up in your connected projects, the sessions using it get their own tab in its info panel, each row ready to open in your DAW * **The assistant reads your open session**: the tracks, samples, plugins, tempo and key in your open project file, plus which of those samples you already own * **Fit the library to your open project**: with a project connected, the **Add filter** menu offers one-click *Fit tempo* (±15 BPM) and *Fit key & scale*, dropped in as chips you can adjust * A project's note now lives on the project itself, and the Notes panel's separate "New Project Note" option is gone ### Embed the free tools on your site The [BPM detector](https://samplevault.ai/tools/bpm-detector) now runs as an embeddable widget you can drop into any blog, forum, or lesson page: copy the snippet from the tool page and paste it where you want it to appear. * Readers drop a file and get the tempo, confidence, and key without leaving your page * Analysis runs in the reader's browser, so nothing is uploaded ### User Experience * **Seek from the player row**: the playing sample grows a slim bar along its bottom edge wherever the row is too narrow to show a waveform, and you can click or drag it to move through a long recording * **Skip back and forward**: `Shift + ←` and `Shift + →` move the playing sample a few seconds a press, and repeat while held * **Compact Database**: a new action in **Settings → App Data** returns disk space the database still holds after you remove folders or samples * **The panel picker, grouped**: opening a new tab lays panels out in sections (Browse & find, Build, Inspect, Workspace) so the one you want is easier to spot * Dragging or copying a sample to your DAW now stops its preview in Sample Vault, so it isn't left looping over your session ### Bug Fixes & Performance * Fixed Locate File discarding a moved sample's tags, favourites, and collections, and deleting its cloud backup * Fixed a folder in the file browser showing a sample count but opening empty * Fixed folder sample counts drifting from the folder's actual contents * Fixed one-shots being marked as loops incorrectly * Fixed silent files reading as full energy in every frequency band * Fixed key and pitch detection being skipped on silent files and very low sub-bass sounds * Fixed setup details (your role, DAW, genres, and challenges) not carrying through to the assistant's instructions * Fixed the Quick Actions "Open in Library" and "Ask Assistant" commands crashing the layout * Fixed finished background jobs being kept for a week after every reprocess, bloating the database on large libraries On macOS, Tape asks for permission to record this computer's audio the first time you switch it on, and needs macOS 14.2 or later. Auto-paste additionally needs Sample Vault enabled under **Privacy & Security → Accessibility**. Samples already in your library keep the loop or one-shot value they have now. To correct a folder in bulk, right-click it and choose **Type → One shot**. To refresh scores on files analysed before this update, use **Reprocess this folder → analysis**. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Library organization, in your control The automatic organization that groups your sounds, tidies your genres and tags, and builds collections now has a home in **Settings → Organization**, with controls to shape what it does and a single place to manage your taxonomy. * **Automatic organization**: turn it on or off, and choose which of its three passes — categories by sound, tidy genres and tags, build collections — run on their own * **House rules**: a plain-language note the AI follows across every pass, like "keep liquid and neurofunk separate" or "no tempo-based collections" * **Your taxonomy, in one place**: categories, genres, tags, creators, and sample packs moved here from Processing — edit them, add your own, and add categories the sound pass can assign * **Lock a genre or tag** so the tidy never merges, renames, or prunes it; anything you create or edit is protected already * **Recent activity**: every automatic pass and manual run is logged here with exactly what changed ### An assistant that remembers you The assistant now carries what matters between sessions — your preferences, the conventions you work by, and what you're in the middle of — so you stop re-explaining yourself every chat. It keeps this current on its own from your recent conversations and library activity, and you can see and edit all of it in **Settings**. * Two kinds of memory: durable facts like your genres, naming conventions, and go-to tempo ranges, plus lighter context about your current work that fades on its own if it stops coming up * Ask it to remember something the moment it matters, or to forget it, without waiting for the next refresh * **See everything it knows** in **Settings → Assistant**, next to your assistant instructions — edit, delete, or switch off any entry, or turn memory off entirely * Confirm or dismiss entries it flags as possibly out of date * A measured read of your library also feeds every chat — the genres you own versus the ones you reach for, your working tempo range, and the searches you keep coming back to * Everything it remembers stays in your library's database on your machine * Cross-session memory comes with Professional Suite, or Lifetime with your own key; the measured library read is on every plan ### User Experience * **More ways to sort**: order the Library by sample length, tempo, loudness, or any sound-profile characteristic * **Mention samples in the assistant**: right-click any selection and choose *Attach to chat*, or add your current selection from the assistant input, to attach specific samples to your next message * The assistant toggle in the sidebar now glows while a request is running and shows a ready dot when a reply lands with the bar hidden * **Stop a reply mid-stream**: the assistant's send button becomes a stop button while it's working, so you can interrupt and send a new message right away * Your assistant instructions moved out of Preferences into the new **Settings → Assistant** page * The **Analysis Speed** control moved into **Settings → Preferences**, and the now-empty Processing tab is gone * **Finer step rates**: Sequencer lanes can now run at 1/64 and dotted rates (dotted 1/4 through dotted 1/32) * In the Library **Add filter** menu, Collection is now a submenu that lists your collections, creates a new collection or live collection inline, and renames or deletes each one * Live collection groups now show the filters behind each group as chips beneath a brighter heading * Ask the assistant to build or show a collection and it opens in the Library scoped to it * **Collect kit sounds to folder…**: from the Sequencer's Export menu, copy every sound in the current kit into one folder you choose — originals stay put * **Copy kit sounds to clipboard**: grab all of the kit's sounds at once and paste them into your DAW or a file window * **Deep tagging summary**: when tagging finishes, a banner shows the creators and sample packs it turned up — click any to filter your library to them * **No more Premium requests**: AI Collections and library auto-organization no longer draw from a monthly allowance — on a paid plan they run under fair use like the assistant and Deep Search * **Honest weak-search note**: when a sound search comes up thin and most of your library isn't tagged in detail, the results now count how many untagged samples actually sound like your matches and point you to describe those folders * **Processing progress at a glance**: opening a samples directory now shows its per-stage progress table up front, with the cloud-backup notice moved into the table under the Uploaded row it applies to * **Reconnect moved files**: a new action finds files you've moved or renamed on disk and reattaches them to everything already known — tags, descriptions, favourites, and collections all stay put ### Bug Fixes & Performance * Fixed filtering by tag or genre stalling the sample browser — the filter counts on a 200k-sample library went from over three minutes to half a second * Fixed the deep-tagged markers on folders and samples staying dark until all background processing finished, long after those folders had been tagged * Fixed a category in the filter panel showing fewer samples than one of its own subcategories * Fixed very large libraries freezing or showing database errors during their first processing pass * Fixed samples dragged into your DAW going missing from the project between sessions Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Loops in the Sequencer Layer always-playing loops under the pad grid: a bassline, a chord bed, a drum break. Each one stretches to your pattern's tempo, and you can swap any of them for a similar loop from your own library. * Drop a loop into a slot and it plays continuously under your beat at your pattern's tempo * Lock a filter to cycle through matching loops, or press ≈ to swap the current loop for one that sounds like it * Mute, solo, volume, and pan per loop * **Build a stack** fills every loop slot with tempo-matched loops from your library in one click, local and free * **Stack around this**: right-click a sample to open the sequencer with that sound as the first loop * **Design with AI** now works the loop lanes too. Ask for a bassline or drum break and it pulls a matching loop from your library * Ask the chat assistant for a kit and it can layer loops underneath, saved into your Sequencer library * Keyed loops follow the transpose control into your target key * Loops are included when you export or render the pattern to WAV ### Filters as chips Every filter you apply in the Library now shows as a chip beneath the search bar, so you can see and edit exactly what's narrowing your results. * **Add filter**: one menu covers tags, genres, category, key, scale, tempo, type, favorites, creator, sample pack, collection, and sound profile * Press F to search the menu and jump to a filter type, or to a specific tag, genre, creator, pack, or collection by name * Click a chip to edit it, its × to clear it, or right-click a tag, genre, or category to flip it to `not:` and drop every sample that carries it * **Suggested filters**: describe a sound and matching filters appear as dashed chips. "bright punchy kick" offers Brightness, Attack, and a Kicks category, to apply one at a time or **Add all** * Structured tokens like `@Cmin` or `#120-140` turn into chips, keeping the search box for plain text * **Shortcuts** with the Library focused: A applies suggested filters, D runs Deep Search, ⌘/Ctrl-B toggles the sidebar, ⌘/Ctrl-S saves a preset, C clears everything ### Deep Search Deep Search now turns your request into a plain search plus editable filter chips, then checks its own results and adjusts the filters until they fit. It handles asks that plain search gets wrong, like "warm pads, nothing cheesy" or "no vocals". * Run it from the **Deep Search** button, ⌘/Ctrl-Enter in the search box, or **Run Deep Search** when a search comes back thin * Its choices land as the same editable chips as any other filter, and your current results stay up until the refined set is ready * **Deep Search from this**: right-click a sample and describe the shift ("darker, more sub, less clicky") to find variations anchored to that sample's measured character * **Shows its work**: a line above the results streams each query it tries, then a one-sentence summary of what it applied or relaxed * Tempo filters match half- and double-time, so an 87 BPM loop still shows up in a 174 BPM search * It reads what a sample sounds like plus its filename and folder, not just its tags, so a mistagged sample still surfaces for the right search ### User Experience * **Add to collection** now starts with **Create new collection**, so you can name one and drop the sample straight in * Step or shuffle through a lane's matches right on the sequencer pads and loops, and hover to see the filter and its match count * Library and File Browser lists keep your scroll position when you switch panel tabs and back * Running a new Library search keeps your current results on screen with a thin progress line, instead of clearing them to a spinner * Tags Sample Vault adds to its own captures and renders (X-Ray, waveform, sequencer) are grouped under a **Sample Vault** filter and left alone when tidying tags and genres * When Sample Vault upgrades your local library on launch, the startup screen now says so, and warns it can take a few minutes on large libraries ### Bug Fixes & Performance * Browsing, filtering, and searching stay fast on very large libraries. The heaviest per-sample data no longer weighs down every list and search * Deep Search and filtered searches now return in about a second on large libraries instead of stalling for tens of seconds, most noticeably with a tag, genre, or tempo filter active * Samples pre-load on hover and press, along with their neighbors, so arrow-key stepping starts each sample instantly * Fixed audio staying silent on macOS after the computer went to sleep and woke back up * Fixed opening a second Sequencer panel blanking out both panels * Fixed switching back to a Library or File Browser tab sometimes showing an empty list or rows at the wrong height * Fixed the Waveform panel showing "No sample loaded" for samples downloaded from a cloud library * Fixed the Library processing status bar staying visible after all processing had finished * Fixed bulk and chat-assistant metadata edits not refreshing what a sample matches on in search * Fixed a large import sometimes slowing to a crawl, with a wildly inflated time-remaining estimate, and the Map view failing to build until the app was restarted * Fixed backups being refused with a "database may be corrupted" error after a recent update, even though the library was intact * **Repair Database** now rebuilds the local database file when an integrity check finds free-space damage, not just drifted indexes, and reclaims the disk space it was holding ## X-Ray Drop any track, loop, or sample into the new X-Ray panel, select any part of it, and find the closest sounds in your own library. It runs on your machine — nothing is uploaded, and it's free. * Open X-Ray as a panel, or right-click any sample and choose **Open in X-Ray** — WAV, MP3, AIFF, or FLAC * Drag across the waveform to select any part — a hit, a phrase, a loop — and the closest matches rank underneath * Regions audition the moment you select them, with a toggle to turn auto-play off * Approve or reject individual matches to steer the ranking without starting over * Save a matched region into your library as a new **X-Ray Captures** source, or copy it to the clipboard * Zoom, fit to width, and reopen recent tracks from the panel ### User Experience * **Spectrogram**: A new panel shows the playing sample as a time–frequency view, sharing the waveform's zoom, scroll, and timeline * **Waveform regions**: Drag across the Waveform panel to select a region that plays on its own, then right-click it to copy it or save it into your library under a new Captures source * **Deep Tagging**: AI indexing is now called Deep Tagging — the metered pass that labels samples with genres, creators, packs, and tags for Deep Search * **MIDI control**: Connect a MIDI controller and play the sequencer's pads with velocity, choosing your device from the transport bar * **MIDI recording**: Press record to capture the pads you perform onto the grid, quantized to each lane's step rate * **Sample folders**: The library now shows the folder each sample came from, with the full clickable path in the sample info panel * Samples with no creator or pack assigned no longer show "Unknown Creator" or "Unknown Sample Pack" * Library search now shows name matches as you type, while sound matches fill in underneath as they're found * In the sequencer, right-click a step to clear it, and the step inspector stays empty until you select a step * Build kit now draws each drum role from across your whole library at random, instead of favoring the alphabetically-first pack * Sequencer pads now hold a minimum size and the grid scrolls sideways when they don't all fit ### Bug Fixes & Performance * Fixed typing a tag, genre, creator, or pack that doesn't exist in your library showing every sample as a match instead of none * Fixed picking certain tags or genres from the search bar making the library flicker endlessly between empty and full results * Fixed samples with accented or unusual characters in their filename failing to analyze and asking you to relocate a file that hadn't moved * Fixed the account menu hiding your Lifetime license when you also own a subscription * Fixed the Starter AI actions counter showing a negative number on accounts granted a larger starter pool * Fixed the library processing indicator staying stuck on "Processing your library" when a sample's file had been moved or deleted * Fixed being sent back to the sign-in screen while still signed in when a background check briefly couldn't reach the server * Fixed mono samples playing silence or the previously-played sound instead of the sample you clicked * Fixed playback starting late while the Waveform or Spectrogram panel was still loading * Build kit now assembles kits much faster on large libraries, with or without a vibe * Randomizing all sequencer lanes now re-rolls every pad at once instead of one at a time ## Faster on large libraries Filtering, finding similar samples, and Deep Search now run roughly 6–8× faster. ### Performance * Filtering by category is now near-instant on large libraries, even as your first filter, instead of scanning the whole library * Finding similar samples and Deep Search both run roughly 6–8× faster * The Similar Samples panel now shows the closest 50 matches first and loads more as you scroll ### Referrals Refer a friend and you both benefit: they save on a Lifetime or annual plan, and you earn a reward when they buy. * **They save**: \$20 off Lifetime, Home Studio annual, or Professional Suite annual on a first purchase. * **You earn**: subscribers get a free month credited to their next invoice; Lifetime owners get an extra device activation. * Referral codes are now available on paid plans, each with a redemption limit. ### User Experience * Category, tag, genre, creator, and pack filters now show how many samples each value covers, updating as you narrow with other filters * Selecting several tags or genres now narrows to samples matching all of them, with a count on each remaining option * Tag and genre groups can now be selected whole in one click, or expanded with an arrow to pick a sub-tag, without opening the group first * A tag or genre group with only one matching tag now shows that tag directly instead of nested * Assistant replies that suggest samples now have an actions row to copy them all, add them to a collection, send them to the Sequencer as a kit, or select them for drag-and-drop and bulk edits * Each sample folder now has a **Remove missing files** action that clears every moved or deleted file in one pass instead of one at a time * Right-clicking a folder in the directory list now opens its full set of actions — scan, retry, reprocess, remove missing files, and delete — without opening it first * The assistant chat box now grows to fit multi-line messages and matches the rest of the app's inputs * The assistant now has a header showing the current chat's title, with controls to start a new chat, browse history, and rename, pin, or delete it * Panel tabs now show each panel's icon, with cleaner selected and hover states * Pressing Ctrl/⌘+F now jumps straight to the library search from anywhere, opening a Library panel first if none is open ### Bug Fixes * Fixed the "Open in Library" quick action doing nothing instead of opening the Library with your search applied * Fixed selected sample chips being nearly indistinguishable from unselected ones in the assistant chat * Fixed the Suggestions panel silently falling back to only recent picks when AI suggestions weren't available instead of explaining why and disabling refresh * Fixed genres and tags you add by hand not joining your genre and tag lists, leaving them out of the filter families and automatic cleanup * Fixed some samples failing analysis because of an unusually low sample rate, such as 8 kHz field recordings * Fixed some 24-bit and floating-point WAV files being incorrectly rejected as corrupt * Fixed some AIFF files, including those Logic exports, being incorrectly rejected as corrupt * Fixed file browser rows flashing white while scrolling on light themes * Fixed the waveform timeline's labels and tick marks being unreadable on light themes ## A library that organizes itself Sample Vault now organizes your library for you — building live collections you describe, correcting categories by how samples sound, and tidying sprawling genre and tag lists. It also keeps verified local backups, so a damaged library is recoverable instead of lost. ### Live collections Describe a collection — "lo-fi drums, warm rhodes, and vinyl textures around 80–90 BPM" — and Sample Vault builds it from your library, organised into labelled groups that keep filling as matching samples arrive. Start one from the collection picker in the Library. * **Grouped browse**: each live collection shows a section per group, with "See all" to open just that group's matches. * **Refine with AI**: adjust a live collection in plain language — "stricter, only F minor" or "add a vinyl textures group". * **Hands-off**: live collections update on their own and stay read-only; manual collections are still yours to fill by hand. * **Built for you**: Sample Vault also assembles a few live collections on its own, shaped by the genres, DAW, and standing instructions in your profile. * Available on paid plans. ### Categories by sound Sample Vault groups samples that sound alike, then uses those groups to fix categories that don't match the sound and to name new ones your library was missing. It runs as part of Organize library. * **Corrects miscategorised samples**: a sample whose category doesn't match how it sounds moves to the one its neighbours share. * **Names new categories**: a distinct kind of sound with no existing home — hardstyle kicks, reverse cymbals, sub drops — becomes its own category. * **Your categories stay put**: categories you've made or renamed by hand are never changed. * Available on paid plans. ### Cleaner genres and tags Sample Vault now keeps your genre and tag lists from sprawling — merging near-duplicates, nesting specific values under broader parents, and dropping tags that only repeat a filter you already have. * **Browse by family**: genres and tags filter as a two-level tree — pick a parent like *trap* or *keys* to match everything beneath it, or open it to choose a single child. * **Your edits stay put**: tags you've created or renamed yourself are never deleted — a cleanup can only nest or relabel them. * Available on paid plans. ### Organize library A single **Settings → Processing** action runs category cleanup, genre and tag tidying, and collection building together — free in the background as your library grows, or on demand for some premium requests. * A status indicator in the title bar shows which step organization is on, whether it started on its own or you ran it. ### Backups and recovery Sample Vault keeps verified local snapshots of your library on a schedule, and when the library can't be read at startup it offers recovery instead of looping on Reload. Manage it under **Settings → App Data**. * **On by default**: choose daily or weekly snapshots and how many to keep. * Each snapshot is verified before it's kept, so a damaged backup is never saved. * **Restore or rebuild**: roll back to your most recent verified snapshot, or re-scan your folders to rebuild the library from the files on disk. ### Interface refresh A pass over the Library and app chrome for one consistent control style that holds up across every theme. * **Consistent filters**: chips, inputs, selects, and toggles in the Library now share one size and surface, with a single emerald highlight for what's selected. * **Cleaner canvas views**: on the Map, Scatter, and Circle views, the fit, isolate, axis, and scope controls moved into a floating panel on the canvas, leaving a slimmer header. * **Quieter title bar**: the workspace, loop, and library-status controls now match the rest of the title bar instead of standing out as boxes. * **Theme polish**: control surfaces and the sidebar now read correctly on light themes and adapt to each theme instead of staying fixed to the dark one. ### User Experience * **Steadier filters**: the Category, Genres, and Tags lists now fill the panel and scroll in place instead of shifting it as options change or you open a sub-group. * **Loading placeholders**: the Category, Genres, and Tags lists show placeholder chips while your library loads, instead of a blank panel. ### Bug Fixes & Performance * Fixed saving an incomplete OpenRouter key showing a generic error instead of telling you the key is too short or missing its `sk-or-` prefix. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.14.4 ### Bug Fixes & Performance * Fixed Ableton Link switching off when you closed and reopened the tempo and key sync popover. * Fixed background sample processing stalling for long stretches while a large library is being indexed. * Background sample details (genres, tags, descriptions) no longer pause while a folder is being AI-indexed, and now keep processing while you're offline. * Sample details now generate roughly 2–3× faster during indexing. * Queueing a folder for AI indexing is faster, by enqueuing them in one pass instead of one at a time. * Fixed the Library status breakdown staying stale after a folder finishes scanning instead of refreshing with the rest of the library. ## v2.14.1 ### Build a kit from chat Ask the Assistant for a kit or a beat and it pulls matching sounds from your library, saves them as a named kit, and drops a button in the chat that opens it in the Sequencer. * **One click to the grid**: the button loads the kit, and an optional starter beat, straight into the Sequencer — opening the panel if it's closed. * **Unsaved work is safe**: if the open Sequencer has edits, it asks before replacing them. ### Manage collections The collection picker in the Library now lets you rename or delete a collection in place, and you can move a whole selection in or out of a collection at once. * **Rename or delete**: hover a collection in the picker for rename and delete actions — deleting drops the collection but leaves the samples in your library. * **Bulk add or remove**: select samples and use **Collections** in the selection menu to add them to a collection, or remove them from one they're all already in. ### User Experience * Switching your update channel now checks for a new version right away instead of waiting for the next automatic check. * The title-bar library status is now a compact icon — its full processing breakdown opens on demand and refreshes from the title-bar refresh button instead of recomputing continuously in the background. * The library now refreshes when new samples first appear, when a scan finishes, and when processing completes, with the title-bar refresh button on hand any time in between — instead of re-querying every few seconds while jobs run. ### Bug Fixes & Performance * A range of performance improvements keep browsing, search, and filtering responsive while large libraries are being scanned and processed. ## The Sequencer (Beta) A pad-grid step sequencer now lives in the dock. Drop your own samples onto pads, program them across polymetric lanes that drift against each other, and hand the result to your DAW — it's built to sketch a groove, not replace the DAW you finish in. The sequencer is in beta. Use the **Feedback** button in the **Sequencer** panel to tell us what's working and what isn't. ### The pad grid Open the **Sequencer** panel and pick a grid from a single row of eight pads up to 8×8. Each pad holds one sample and maps to a MIDI note for later. * **Fill a pad**: pick from your library, drop a sample on it, or let it choose — click a pad to audition it. * **Per pad**: mute, solo, and lock. ### Polymetric step lanes Every pad gets its own lane with its own length and rate, so lanes fall out of sync and only line back up after a long cycle. * **Length and rate per pad**: 1–32 steps at anything from 1/4 down to 1/32, plus triplets. * **Velocity per step**: drag or scroll a step to set how hard it hits. * **Per-step pitch**: tune individual steps, with the resulting note shown against the pad's key. * **Snap to key**: lock step pitches to the detected key and scale so nothing clashes. * **Choke groups**: let pads cut each other off, the way a closed hat silences an open one. ### Sourcing from your library The sequencer leans on what Sample Vault already knows about your sounds. * **Swap for similar**: replace a pad with a sample that sounds like it. * **Build a kit**: fill the unlocked pads with a matching set of sounds from your library. * **Design with AI**: describe a vibe and the designer picks the kit, programs the pattern, and sets the tempo and swing to match. * **Design one lane**: point the same AI at a single pad to swap its sound, rework its rhythm, or both. * **Shared tempo and swing**, following Ableton Link when a session is running. ### Export to your DAW When the groove is right, take it with you. * **MIDI clip** of the whole pattern, each pad on its own note, plus a manifest mapping every note to its sample. * **Audio render** to WAV, clamped for polymetric patterns whose loop runs long. * **Save as a sample**: render the pattern straight into your library — analyzed on the spot and tagged as a loop at the pattern's tempo, searchable immediately. * **Copy to clipboard**: send the MIDI clip or the WAV render straight to the clipboard to paste into your DAW. Sequencer playback and kit suggestions run on your machine and are free. Patterns save automatically as you work. ### Plans & Usage Deep Search, the assistant chat, the Sequencer Designer, and library suggestions now share one **AI actions** allowance instead of separate search and chat grants — heavier actions cost more from the same pool. Free accounts get a one-time starter grant; paid plans stay effectively unlimited under fair use. The Premium-requests pool (Smart chat, AI collections) and the indexing coverage pool are unchanged. See [Plans & Usage](/reference/plans-usage#starter-pools). ### Bug Fixes & Performance * Fixed the assistant failing with a cryptic error when your session expired mid-conversation, instead of telling you to sign in again Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.12.0 ### User Experience * **Library status panel**: The status popover mirrors the per-folder view — All / Ready / Working / Failed counts, a Details toggle for the full processing pipeline, and a live processing rate — with the activity feed now behind a View full activity button. * Reprocessing, retrying failed samples, and deleting a directory now use the same cancellable progress toast as other bulk actions, on a single folder or every directory at once * Similar Samples now offers a **Classify now** button when a sample hasn't been analyzed yet, so you can process it on the spot instead of waiting for background classification ### Bug Fixes & Performance * Fixed large libraries slowing browsing and search to a crawl the longer indexing ran * Fixed the Library's category filter crashing the view when a sample carried a malformed category value * Fixed your sample directories disappearing from settings while working offline on a device-activated license * Fixed CPU spikes and stalled processing when changing the Analysis Speed setting while a library was still indexing * **Faster library scans**: Adding a folder writes new samples up to 8× faster on large libraries, so a scan that used to crawl for an hour finishes in minutes. * **Responsive during scans**: Browsing, searching, and playback stay smooth while a large folder scans or indexes in the background, instead of stalling for seconds at a time. * **Faster search-readiness**: New samples become searchable by sound several times faster after a scan, and faster still once the scan's other work has finished. * Fixed the app freezing for up to a minute, or edits and saves failing, when heavy scanning and indexing ran at the same time. * Fixed your remaining Deep Search count lagging behind actual usage after a search, AI suggestions, or a sequencer design. * Fixed search tokens being too pale to read on light themes across the library search, sequencer query, and command palette inputs. ## See Your Whole Library The Sample Browser is now the **Library**, and it lays your collection out four ways: a list, a map of similar sounds, a circle of fifths, or a scatter plot. Search collapses to one bar that matches filename and sound together, with an optional cloud AI pass on top. This release also adds app themes and Google Drive backup. ### Four views of the Library The renamed **Library** panel keeps every workflow and adds a view switcher; existing dock layouts migrate automatically, and each panel remembers its own view, search, and filters. * **List** — the familiar browser. * **Map** — samples that sound alike sit close together, coloured by category, thinning to representatives when you zoom out; scope it to one category or the whole library. * **Circle** — the circle of fifths, majors outside and relative minors inside; click a key to filter, and pick what distance from centre means (tempo, brightness, attack, and the rest). * **Scatter** — every sample plotted on two characteristics you choose, axes swappable in a click. * **Shared across views** — click plays, drag drops into your DAW, filters and search dim non-matches (**Isolate** hides them), and **Fit** / **Regenerate** reframe or rebuild the layout. ### Deep Search The search bar lost its modes. Type, and it matches by name and by how samples sound at once, free and offline. The new **Deep Search** button runs a cloud AI pass for sharper matches, and falls back to the free search when your allowance runs out. * **Ideas to try** — suggested searches under the bar surface sounds worth playing next, and **Surprise me** leans on the corners of your library you rarely open. * **Relevance by default** — results lead with the closest name and sound matches instead of newest first. ### Themes Sample Vault now ships more than one look. Pick a theme in **Settings → Display** and the whole app follows. Your choice sticks between sessions. * **Neutral** — Midnight (the default), Graphite, True Black, Dim, Ember, Paper, and Daylight. * **Accessibility** — Contrast and Contrast Light for low-vision use or bright rooms. * **Studio classics** — Wavetable, 808, Session, Console, Pocket, and Screen, each with one bold accent. ### Bring Your Own Google Drive Back up and sync new uploads through your own Google Drive. Connect under **Settings → Connections → Cloud storage**, and files land in a top-level `Sample Vault/` folder mirroring your source folders. * **Three-way default** — switch new uploads between the built-in cloud, your Dropbox, and your Google Drive without disconnecting any. * **Your own cloud needs no subscription** — only the built-in cloud requires a Home Studio or Professional Suite plan. ### Also New * **Copy the playing sample with Ctrl/Cmd+C** — drops its file on your clipboard ready to paste into a DAW, while highlighted text still copies as text. * **Redesigned Sample Directories settings** — a cleaner master/detail layout with per-directory processing detail. * **Tidier Dropbox folders** — new uploads land directly under `/Apps/Sample Vault/` mirroring your source folders. ### Performance and Bug Fixes * **Lower-priority analysis** — indexing runs below your other apps, so a large library processing in the background no longer steals CPU from your DAW. * **Better-spread characteristics** — the sound traits behind the views and filters now use their full range, so sounds separate more clearly. * Fixed the app occasionally starting up unable to load your library until you reloaded the window. * Fixed AI suggestions coming up empty on machines without on-device sound matching. * Fixed analysis occasionally stalling a few samples short of 100%. * Fixed the filter sidebar reopening every time you switched between layouts that reuse the Library panel. * Fixed downloading a sample in the web app opening it in the browser instead of saving it to disk. * Fixed the web app intermittently failing a sample copy or download instead of refreshing your sign-in session. * Fixed clearing completed jobs from the activity log failing on large libraries with a big backlog. * Fixed the right-click menu not opening on macOS when Control-clicking instead of using a two-button mouse. * Fixed the folder shortcuts in Settings → App Data failing to open in Finder on macOS. * Fixed dragging a file into the app window on macOS replacing the app with that file. * Fixed uncategorized samples being left out of the Library's category filter instead of appearing under Unknown. * Fixed loosely-named vocal samples and other one-shots staying uncategorized instead of landing in the right family. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.12.0 ### Directory health and failure triage Each sample directory's settings view is now a triage console: see what's ready, still processing, or failed at a glance, and fix problems without leaving the page. * **Three states** — every directory splits into Ready, Working, and Failed, so a library that's still processing reads differently from one that's stuck. * **Failures grouped by cause** — failures are grouped (file was locked, unsupported format, file moved or deleted) and split into the ones a retry can fix and the ones it can't. * **Retry failed** — one button re-runs everything recoverable, while corrupt or missing files are called out with what to do instead of being retried forever. * **Live progress** — the Working count shows how fast samples are clearing and roughly how long is left, with a shortcut to the processing speed controls. * **Find a file fast** — the sample list filters by name and stays smooth at tens of thousands of files, with multi-select for bulk retry, reveal, or delete. * **Reprocess on demand** — re-run analysis, classification, enrichment, or cloud upload for a whole directory or your entire library from the directory menu, without deleting anything. ### Cancel any bulk action Folder- and selection-wide actions — setting key, tempo, or genres, analysing, classifying, enriching, uploading, AI-indexing, deleting, rescanning — now show live progress with a Cancel button. * **Stops where it is** — cancelling keeps whatever already finished instead of rolling the whole batch back. * **Consistent across the app** — directory menus, multi-select edits, the Activity Log's retry and clean-up, and reanalysis all behave the same way. * **Steady on big libraries** — bulk edits over tens of thousands of samples stay responsive instead of stalling, especially on Windows. ### User Experience * **Now playing in the Map, Circle, and Scatter views** — clicking a dot brings up a player at the bottom of the view with the sample's name, key, tempo, and waveform, plus play/pause, scrub, copy, drag, and favorite, so you can tell what's playing without leaving the layout. * Moved **Delete directory** into an overflow menu so the recovery actions, not the destructive one, lead the directory view. * Renamed **Rescan** to **Scan for new** and added a heads-up before scanning very large folders, since scanning only picks up newly added files — it won't retry failures or reprocess existing samples. ### Bug Fixes & Performance * Fixed some MP3 files being marked corrupt and skipped during analysis even though they play fine * Fixed folder and multi-select edit dialogs freezing — unable to close or start another edit — while a previous bulk change was still applying to a large library * Fixed automatic genre, tag, creator, and sample-pack matching skipping samples whose names or folders were close to a known label but not an exact match ## Duplicate Detection and Playback Cleanup Sample Vault now recognises audio duplicates and collapses them so each sound shows up once. A new non-destructive playback chain evens out loudness, trims silence, removes DC offset, and de-clicks edges. Originals on disk are never touched. ### Duplicate Detection Every analysed sample gets a signature based on what it sounds like. Matching samples are grouped, one copy shows in the browser, search, and similar results. The rest are hidden, never deleted. The same sound at WAV / AIFF / FLAC, 16- / 24-bit, 44.1 / 48 kHz, or normalised to a different level all count as one. **Settings → Duplicates** shows what's collapsed and the controls: * **Collapse toggle** — turn off to see every copy again. * **Match strictness** — *Exact* through *Looser* (near-identical, lossy re-encodes, light edits). Starts at Exact. * **Scan now** — re-check on demand. Also runs after every directory scan. * **Review** — flip copies between *kept* and *hidden* per group. * **Move to Trash** — send hidden copies to your system Trash. The kept copy is untouched. Samples added before this update need to be reanalysed to detect duplicates correctly. ### Playback cleanup Every newly scanned sample now carries peak (dBFS), integrated loudness (LUFS), RMS, and silence-boundary measurements. **Settings → Preferences → Playback** (and the popover behind the title-bar volume button) turns these into a non-destructive chain: * **Normalize playback loudness** — match each sample to a target true-peak (default −1 dBFS, range −6 to 0). * **Trim leading / trailing silence** — skip auto-detected dead air. Head and tail toggle independently. * **Remove DC offset** — subtract DC bias before playback. On by default. * **Boundary declick** — 1–50 ms symmetric fade at sample edges to kill ticks. ### Render on drag / copy Turn on **Render samples as heard when dragging or copying** to bake the active chain into the file your DAW receives. Off by default. Rendered copies preserve the original sample rate and bit depth. ### Bring Your Own Dropbox Route new uploads through your own Dropbox account. Connect from **Settings → Connections → Cloud storage → Connect Dropbox** — files land under `/Apps/Sample Vault/`, mirroring your source folders. A **Default upload destination** card flips between built-in cloud and Dropbox without disconnecting either. Cloud uploads still require Home Studio or Professional Suite. See [Cloud Storage](/reference/cloud-storage). ### Bring Your Own AI key Paste an OpenRouter API key in **Settings → AI** to route AI calls through your account. * **Key only** — every call uses your key; no Sample Vault pool is touched. * **Pools first, then key** — pools drain first, your key picks up when they're empty. Also the only way to keep Smart chat, AI Collections, Prompt search, and AI indexing working on Demo Room or One-Time License after starter grants run out. See [Plans & Usage → BYOK](/reference/plans-usage#bring-your-own-key-byok). ### Delete to Trash Sample deletion can now move the underlying audio files to your system Trash instead of just removing them from the library. Available on single, multi-select, and directory deletes. Off by default. ### Activity Log management Right-click any job to delete or requeue it. Right-click a grouped row (e.g. *"12,000 enrichment jobs pending"*) to apply the action to every job in the group, with a confirmation showing the count. Requeue is offered only for failed or completed jobs. Bulk operations stream a live progress toast. ### User Experience * **Analyze samples** on a folder no longer pops a confirmation — it queues right away and reports in a toast, matching Classify, Enrich, Upload, and Index. * **Scans become visible immediately** — directories start populating the browser as the first batch lands, not after a minute's wait. * **Analysis and enrichment results land sooner** — finished runs surface in the browser instantly. * **Title bar refresh icon** appears while background changes are pending; click to pull them through. * **Tempo Lock / Key Lock** now show retuned BPM and key on every row in the sample browser, not just the playing row. * **Audition volume slider** moved to **Settings → Preferences → Playback** alongside the cleanup toggles. * **Add sample directory** disables Upload and AI Index toggles when your plan or sign-in state can't run them, with a one-line note explaining what's needed. * Right-click menus now scroll when taller than the window and no longer get clipped at screen edges. ### Browser demo * The browser demo now opens straight into the full app — workspaces, tempo and key sync, play/pause, auto-loop, and the playback options popover all work against the demo library. Install lives in the title bar. ### Bug Fixes & Performance * Fixed the browser demo blocking anonymous visitors with a "Subscription required" screen on first load. * Fixed reanalysis wiping tags and overwriting key, tempo, or loop values you'd set by hand. * Fixed AI enrichment replacing genres you'd set by hand instead of merging with the folder's detected ones. * Fixed device activation not refreshing the rest of the app — the API Keys page and profile menu sometimes kept showing "activate this device" until the next refetch. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.8.2 ### Bug Fixes & Performance * Fixed right-click context menus not opening in some places in the app. ## v2.8.1 ### Bug Fixes & Performance * Fixed scanning a new sample directory silently dropping every file — the scan reported "done" but added nothing to the library. ## Sample Vault in the Browser Sign in at [samplevault.ai](https://samplevault.ai), click **Open app**, your library opens in any browser tab. ### Browser Access * Sample Browser, filters, Sound Profile, search modes, Similar Samples, AI Assistant, Quick Actions all work the same. Edits sync back to your other devices. * Only cloud-uploaded samples are visible. Local files stay hidden. * Local-only features (scanning, analysis, indexing, Ableton Link, drag-from-disk) show a "desktop app required" prompt. * Globe pill in the title bar marks browser mode. Included on **Home Studio** and **Professional Suite**. See [Browser Access](/reference/browser-access). ### Suggestions Panel A new dockable panel that surfaces samples based on your recent activity, the sample playing, and your project's tempo and key. * **Recent moves**: anchored on your last drag or copy. Local, no network call. * **AI-picked sections** (2–4 per refresh) with titles like "A bass for the drop" — uses semantic search, similarity, or hard filters per section. * Refreshes on new prompt search, drag, or project switch. ### User Experience * **Calmer chrome** across the app: softer filter borders, ghost play buttons that fill only when playing, mono loop / one-shot badges. * **Type filter** is now two toggles (Loops, One-shots). **Sort filter** collapsed to a single dropdown. * **Dock tabs** muted at rest. × hidden until hover, "+ New Tab" only visible on active panels. * **Quick Actions suggestions strip** under the search input. Tab / Shift + Tab to cycle, Enter to insert. * **Settings → Keyboard Shortcuts** lists the full set, including arrow-key playback (`↑`/`↓`, `Space`, `Shift + Space`). * **Retry failed** button in Activity Log requeues failures with fresh attempts. Respects the type filter; batched for large queues. * **First-run setup** jumps straight into **Add sample directory** after you click **Open Sample Vault**, so you can point at your first folder immediately. ### Bug Fixes & Performance * Fixed F5, Ctrl/Cmd + R, and right-click → Reload resetting the window mid-session. * Fixed sign-in refresh loops in Firefox. * Fixed "Sign in" from the marketing site landing on Account with an error after the browser demo. * Fixed a single failed database query cascading into errors across sample browser, search, and indexing until app restart. * Fixed Quick Actions palette ignoring `category:`, `&creator`, `+pack`, `*`, characteristic ranges, `collection:`, and `sort:` tokens. * Fixed Quick Actions palette "Open in Sample Browser" and "Prompt Search" actions doing nothing. * Fixed phantom hover square on Collection, Creator, and Sample Pack filters when empty. * Fixed the app asking you to sign in every day. * Fixed AI directory indexing failing when the AI model was temporarily unavailable — it now falls back to an alternate model. * Fixed library coverage not being refunded when AI directory indexing failed. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.7.1 ### Free in-browser tools Two new tools at samplevault.ai/tools — same audio engine as the desktop app, runs locally, no upload, no sign-up: * **[Pitch Shifter](https://samplevault.ai/tools/pitch-shifter)** — transpose any audio file by ±12 semitones with the tempo unchanged. Preview both versions, download the result as WAV. * **[Time Stretcher & BPM Changer](https://samplevault.ai/tools/time-stretcher)** — drop a file, source BPM auto-detected from the filename or by analysing the audio. Set the target BPM (or pull a rate slider directly), preview, and download the matched WAV. Pitch unchanged. ### Bug Fixes & Performance * Fixed database connection getting stuck on an unexpected error and all operations failing. * Fixed dragging a sample on macOS sometimes opening the audio file inside the app and forcing a restart. ## Multi-Select Everywhere Selection works consistently across every place you see samples — Sample Browser, Prompt Search, File Browser, Similar Samples, and the assistant chat. Pick a batch in one panel, add to it from another, drag the whole set into your DAW, or bulk-edit through one consistent menu. The same shortcuts work everywhere: **Cmd/Ctrl + click** toggles, **Shift + click** ranges, **Cmd/Ctrl + A** selects every sample currently loaded. In the Sample Browser, scroll to load more and press Cmd/Ctrl + A again to extend. In the assistant chat, Shift + click ranges within a single message, and with nothing focused Cmd/Ctrl + A falls through to text selection as you'd expect. ### Repair Database A new "Repair Database" button in Settings → App Data runs an integrity check and rebuilds drifted indexes. Use it when scans, deletes, or batch operations stall. Safe to run any time, no data loss. ### Free in-browser tools Three new tools at samplevault.ai/tools — same DSP as the desktop app, runs locally: * **[BPM Detector](https://samplevault.ai/tools/bpm-detector)** — tempo, detection confidence, half/double alternatives, loop check. * **[Mono-Compatibility Checker](https://samplevault.ai/tools/mono-compatibility)** — per-band phase incompatibility (sub-bass through highs) so you can spot what collapses on club PAs, phone speakers, or vinyl. * **[Camelot Wheel Converter](https://samplevault.ai/tools/camelot-key)** — pick a key (or drop a track to detect) and see the three harmonically-compatible neighbours for DJ transitions. ### Bug Fixes & Performance * Fixed "Upload to cloud" actions queueing uploads that would never succeed for Demo Room users and over-quota subscribers. Opens the upgrade dialog instead. * Fixed the delete confirmation referring to non-refundable "credits" instead of library coverage. * Fixed the browser demo getting stuck on "Downloading your library samples". * Fixed selecting a top-level category (Instruments, Drums, FX, Synth, Vocals, Other) in the sample browser sidebar clearing itself instead of filtering. ## Help & Feedback in One Place One front door for everything help-related. A new sidebar icon opens a single dialog covering documentation, changelog, keyboard shortcuts, bug reports, feature requests, messages, and the Discord invite. The right-click menu, profile menu, error screen, and the new `Ctrl/Cmd+Shift+/` shortcut all open the same dialog. ### Send feedback without leaving the app * **Report a bug** attaches diagnostics automatically: app version, OS, recent errors, and up to the last 10 MB of the local log file. A summary line shows what's being sent, and you can switch the toggle off if you'd rather send nothing. * **Request a feature** captures a short title and description. * **Send a message** is for everything else — questions, comments, things that don't fit a category. * The error screen now has an "Add details about what happened" button that opens the bug form pre-filled with the error you just hit. ### User Experience * The "Update available" notification now pops up automatically as soon as a new version finishes downloading, instead of waiting for you to click the sidebar button. * Both the "Update available" and post-update "Updated to vX.Y.Z" notifications now show the release notes for that specific version when available. * Search bar suggestions now surface sample-type and category filters from plain typing. Type `loop` or `one-shot` to see the matching type filter; type a category name like `kick`, `snare`, or `drums` to see matching folders from your library — no need to remember the `!loop` / `category:` syntax. ### Bug Fixes & Performance * Fixed the Add Directory dialog flashing back to the top of the subfolder list while other samples were being processed in the background, making it hard to scroll through and select subfolders. * Fixed deleting, bulk-editing, or resetting libraries with more than \~900 samples taking 2-3 minutes. The same operations now finish in under a second. * Fixed macOS metadata sidecars (`._` files that macOS writes next to real audio when copying to exFAT/NTFS drives) being treated as audio samples. They no longer get scanned, no longer appear in the library, and no longer fail to play with a "file path is not permitted" error. If you're already seeing `._`-prefixed entries in your library from before this fix, remove and re-add the affected directory, or ask the AI assistant to clean them up for you. ## The Master Search Bar ### Sample Browser Enhancements Every filter in the sample browser now runs through the search bar. Tags, genres, creators, sample packs, categories, collections, sort order, sample type, favorites, BPM, key, and all 16 sound profile characteristics are each a typeable, color-coded chip. Pick from the suggestion dropdown to learn the syntax, click a chip to remove it, or type the tokens directly if you already know them. The chip strip above the results is gone, and the toolbar that replaces it stays visible whether or not the filter sidebar is collapsed. #### New tokens Eight short prefixes plus `field:value` long forms cover every filterable dimension. | Dimension | Short | Long | Example | | - | - | - | - | | Tag | `%bass` | `tag:bass` | `%dark %"sci fi"` | | Genre | `$dnb` | `genre:dnb` | `$"drum and bass"` | | Tempo | `#120` | `bpm:120` | `#90-140` `#100-` | | Key + scale | `@Cmin` | `key:Cmin` | `@F#min` | | Sample type | `!loop` | `type:loop`, `is:loop` | `!oneshot` | | Favorite | `*` | `is:favorite` | `*` | | Creator | `&adam` | `creator:adam` | `&"Adam Smith"` | | Sample pack | `+cymatics` | `pack:cymatics` | `+"Cymatics Hyperbits"` | | Category | — | `category:drums/kicks` | `category:bass` | | Collection | — | `collection:wip` | `collection:"Work in Progress"` | | Sort | — | `sort:newest` | `sort:tempo-asc` | | Sound profile (×16) | — | `atk:30-80`, `bright:0-50`, etc. | `stereo:80-100` | Sound profile aliases include `atk` (attack), `bright` (brightness), `bass`, `sub`, `mids`, `highs`, `lomids`, `stereo`, `perc` (percussiveness), `dyn` (dynamic range), `pitch`, `rhythm`, `tonality`, `noise`, `complexity`, `temporal`. Range syntax matches BPM: `atk:30-80` (between), `atk:30-` (at least), `atk:-80` (at most), `atk:50` (exact). #### Smart suggestions dropdown * Type two characters and the dropdown opens with matching values across all dimensions, grouped by section. Pick one to insert the right token automatically. * A "Filters" section at the top shows field shortcuts (`creator:`, `pack:`, `collection:`, etc.) when your prefix matches one. You pick up the grammar by using it. * After committing to a dimension (typing `&` for creator, `category:` for category, etc.) the dropdown immediately shows top values without the two-character gate. * The top of the dropdown also offers shortcuts to run your current query in the *other* search modes. For example, while typing in Find mode you can jump straight to Prompt or Instant search without manually toggling modes first. * Suggestions are validated against your library: typing a partial value like `&ad` no longer applies a partial-name filter that zeroes out results. * Keyboard: arrow keys to move, `Tab` to autocomplete, `Enter` to apply, `Esc` to close. #### Chip interactions * Click any chip in the search bar to remove just that filter. On hover, the chip shows a red strike-through and × icon previewing the removal. * The **×** button on the right side of the input clears every filter at once. * Chips track the input's scroll position now, so typing into a long query keeps the caret visible and the chips aligned with what you're editing. #### Toolbar reshuffle The chip strip above the results is gone. A single toolbar replaces it with the search mode toggle, the search bar, favorites, presets, and the sidebar collapse button. The search bar stays visible regardless of sidebar state, so collapsing the sidebar gives you full-width browsing without losing access to filtering. The mode toggle now uses a proper segmented control with a distinct active state. Buttons in the toolbar grow text labels at wider widths (≥ 520px) and collapse to icons at narrower widths. #### Other improvements * Tag and genre filter sections in the sidebar now show every value in your library, sorted alphabetically. Previously the list was capped at 50 and ordered by frequency, so less-common tags were either missing or appeared in a confusing order after expanding "+more". * Search bar font is slightly tighter (13px) so more chips fit on screen. * Smarter placeholder text in Find mode shows token examples right in the input. * Empty `category:` (or any other field with no value yet) is no longer treated as a filename search. Partially-typed tokens are correctly ignored until you finish them. * Fixed typing into a search bar that already has many tokens not scrolling the chip overlay to follow the caret. Previously the visible text stayed static while you typed beyond the right edge. * Help dialog (the `?` button) is now wider with a proper table layout for tokens and modes, and uses higher-contrast text throughout. ### Performance * Folders in the file browser tree now preload their contents on hover, so expanding feels instant. * The tag and genre filter rail opens roughly 7× faster on large libraries, sub-second instead of multiple seconds. * Background indexing no longer stalls the sample browser or filter sidebar for seconds at a time on Windows. * Processing status and the System Status panel refresh roughly 6× faster. * Browsing large libraries is much faster on Windows. The main sample list now loads in milliseconds instead of seconds, even at 1M+ samples. ### Bug Fixes * Fixed stacking multiple tags or genres widening the matching samples (OR) instead of narrowing them (AND). The remaining filter chips correctly shrink as you add constraints. * Fixed assistant chats getting stuck on a tool call that never finishes when the connection drops mid-request. The interrupted tool now shows as "interrupted" with a one-click **Continue** button that picks the conversation back up from where it left off. ## New pricing: No more credits Clearer pricing, faster indexing, and a long list of UX cleanup. ### No more credits "Credits" are gone. Library size is now **unlimited on every plan** — only the AI indexing pass that detects genre, mood, creator, and tags is metered. Two pools replace the single credit balance: * **Premium requests** — Smart chat and AI collections. * **AI indexing** — samples tagged per month. Monthly limits per plan: * **Home Studio**: 50,000 indexed samples, 100 Premium requests. * **Professional Suite**: 200,000 indexed samples, 400 Premium requests. * **Demo Room** (free, one-time grants): 100 AI searches, 30 chats, 5,000 indexed samples. ### Library Status and Usage * A dot on your profile avatar warns when a usage pool is running low. Open the menu to see which. * Billing shows clear per-pool meters with reset dates instead of a single opaque credit number. * The Library status dialog opens with a one-line verdict — *"Analyzing 1,248 of 3,500 samples"*, *"All caught up"*, or *"Nothing's running"*. * Adding a folder that exceeds your remaining coverage now tells you how many samples will be queued vs. skipped, instead of blocking the whole folder. ### Indexing Performance * AI directory indexing is significantly cheaper and faster — sibling folders are tagged together. * CPU stays saturated through long indexing runs, especially on Windows. RAM stays flat instead of climbing. * The **Analysis speed** presets (Battery / Balanced / Fast / Max) now actually cap CPU usage. Previously every preset pegged multi-core machines at \~100%. * Large DSP speedup on Windows. * Long files (DJ mixes, full albums, multi-hour recordings) are skipped instead of loaded into memory and discarded. * The sample browser and filter sidebar no longer stall for seconds at a time during background indexing. Windows and Linux x86\_64 builds now require a CPU with AVX2 — Intel Haswell (2013+), AMD Excavator (2015+), or newer. Older CPUs need to stay on v2.4.1. ### Sample Browser * **Continuous scroll** replaces Previous/Next pagination. Auditioning through long searches no longer breaks every 25 samples. * Scrolling is noticeably smoother on large libraries. ### User Experience * Sign-in and startup splash screens now show window controls, so you can move or close the window before the app finishes loading. * While offline, the title bar shows a single amber **Offline · Sign in** pill instead of a duplicate sign-in button. ### Bug Fixes * Fixed default workspace layouts being duplicated on launch when the initial cloud sync was slow. Existing duplicates clean up automatically. * Fixed directory indexing being partially charged or silently completing with empty tags when coverage ran out mid-batch. Coverage is now refunded in full, and jobs fail upfront with a clear message. * Fixed assistant messages scrolling out of view while tool calls were running. Your last message now stays pinned near the top. * Fixed sign-in showing a generic error toast when requesting a verification code too quickly. You now see an inline countdown. * Fixed multi-word search tokens (e.g. `$drum and bass`, `%hip hop`) dropping everything after the first word. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Faster Indexing, Heavier Right-Click ### Parallel Indexing Pipeline Local audio analysis (tempo, key, spectral, stereo) and AI classification now run as independent workers with their own thread pools. Tempo, key, and waveforms show up much sooner on a fresh scan, and tagging no longer throttles DSP work. * **Richer System Status panel**: processing now breaks into six stages — Total, Analyzed, Classified, Uploaded, Enriched, and Embedded — each with a tooltip explaining what it unlocks (tempo/key search, similarity, sync, genre/mood filters, prompt search) * **Classify samples on demand**: right-click a sample or folder and pick **Classify samples** to re-run AI classification without redoing local analysis ### Right-Click, Edit Anything The sample right-click menu now has quick editors for **Key**, **Scale**, **Tempo**, **Creator**, **Sample pack**, **Type** (loop/one-shot), and **Category**. Creator and Sample pack open a searchable picker; Tempo suggests your library's most common BPMs as chips; Category mirrors the full tree. * **Batch-edit directories**: right-click any folder in the File Browser to set the same properties on every sample under it, with live progress on large folders * **Tags and genres, inline**: dedicated Tags and Genres dialogs on samples, and Add/Remove Tags/Genres on directories. Remove dialogs only list values that actually exist under that folder * **Rename or delete tags and genres library-wide**: right-click any tag or genre chip in the filter rail or Sample Info pane to sweep or strip it from every sample at once ### Per-Folder Upload and Indexing Control The Add Directory dialog now shows your sample folders as a tree with per-directory checkboxes for cloud upload and AI indexing. Live directory count, file sizes, and credit estimates update as you toggle folders, so you can see exactly what a scan will cost before committing. * **Simplified credits**: removed the Safety Reserve setting — your whole balance is now available for both AI features and background processing ### Also New * **Smarter search for poorly-named samples**: each sample now inherits its parent folder's AI description when generating its embedding, so a file like `01.wav` inside a "Vintage 70s funk drum breaks" folder matches searches for that vibe. Applies to newly enriched samples ### Bug Fixes & Performance * **Faster local analysis on laptops**: background analysis no longer oversubscribes CPU threads when multiple classifier sessions run in parallel; audio resampling during tagging is also \~3× faster. Biggest gains on 4–8 core machines * **Smoother indexing throughput**: results now save as each file finishes instead of waiting for the whole batch, so one slow file no longer stalls the rest * Fixed the assistant sometimes failing to send a message and leaving behind an empty chat that couldn't be reopened after refreshing * Fixed re-indexing a folder while a previous index was still running queueing duplicate jobs for the same directories * Fixed dragging a sample and dropping quickly occasionally loading the audio file in the app window and forcing a restart (mostly on macOS) * Fixed similar-samples search breaking after downloading a bundled library. Existing libraries are repaired automatically on next launch Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Unified Search: Find, Prompt, and Instant The [Sample Browser](/core-features/sample-browser) now has three search modes in one place — pick the one that fits how you're searching right now. ### Find · Prompt · Instant A mode toggle above the search bar lets you switch between three ways to search. Press `Tab` while typing to cycle modes. * **Find** (default): filters by filename as you type. Free, instant, fully offline. * **Prompt**: describe any sound in plain English — AI returns the best matches even when filenames don't help. Costs 5 credits per search. * **Instant**: free on-device matching using a local model. Fast and offline, but less accurate than Prompt. All three modes support the same search tokens (`#120 @Cmin !loop %dark $techno`) for filtering by tempo, key, type, tag, and genre. In Prompt and Instant, tokens act as hard filters while the rest of your query drives the AI match. The search bar and sidebar filters now stay in sync — type `@Cmin` and the Key dropdown updates, or pick a key from the dropdown and the search bar updates to match. ### Analysis Speed A new single slider in **Settings → Processing** replaces the old separate thread and model controls. Four presets — Battery, Balanced, Fast, Max — with a live RAM and CPU estimate so you know what to expect. Also available in the **Add Directory** dialog and the **System Status** panel while analysis is running. ### User Experience * **Favorites** moved to the top of the filter sidebar for quicker access * Active filters are now visually highlighted so you can see at a glance what's narrowing your results * Search help button (?) opens a quick reference with mode comparison and token syntax * New **Help** button in the sidebar footer links to the documentation ### Bug Fixes & Performance * Fixed the update notification showing "Update & Restart" before the download finished * Fixed sign-in and sign-up links dropping the redirect parameter after authenticating * Fixed the Getting Started checklist briefly flashing on launch for users who had already completed it Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.2.2 ### User Experience * **Font picker**: Choose your preferred UI font in Settings → Preferences. Ships with Space Grotesk (default), Inter, Geist Sans, Plus Jakarta Sans, Sora, Exo 2, Oxanium, and JetBrains Mono ### Bug Fixes & Performance * Significantly faster sample and directory deletion, especially for large libraries * Fixed scanning directories failing with "An unexpected error has occurred" when a filename contained unusual control characters * Directory scans that were interrupted by closing the app now automatically resume on restart ## Ask Your Library Anything The AI assistant can now query and edit your entire sample library directly. Ask analytical questions, run similarity searches with filters, and make bulk changes — all with inline approval before anything is written. AI assistant running a library query with inline approval panel ### Assistant SQL Tools * **Ask questions about your library**: "how many kicks between 60–80 BPM?", "which genre has the most samples?", "show me favorited samples I haven't played in a month" * **Similarity search with filters**: "samples that sound like this one but only lofi, under 80 BPM, in F minor" — combines vector similarity with any filter you describe * **Make bulk changes**: "archive everything tagged lofi that isn't favorited", "remove the 808 tag from all kicks" — the assistant proposes the change and you approve it inline before anything happens * **Safe by default**: Destructive statements are blocked automatically, and every write shows a confirmation card with a plain-English summary before executing The assistant can now answer questions and act on your library in ways we haven't pre-built features for — any question you can describe, it can query. ### User Experience * Prompt Search now works in the browser demo * Web account area consolidated into a single page — plan, usage, referral, and settings in one place * Email preferences now sync between desktop and web * Assistant chat messages pin your question to the top so you read the response from the beginning * Tool steps in the assistant are now compact — in-progress steps show a spinner, completed steps collapse to a single line * Notes editor is tighter and more compact * The app uses custom scrollbars in more places to provide a more native feel ### Bug Fixes & Performance * Fixed web account page throwing errors when adjusting preferences * Fixed global shortcuts not being recognized * Fixed demo video not playing on some browsers * Fixed email verification sign-in bouncing back to the sign-in page Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## v2.0.8 ### User Experience * "Check for Updates" in the user menu now opens the update dialog directly when a new version is found ### Bug Fixes & Performance * Fixed sign-in on macOS silently returning to the sign-in screen after a successful browser auth * Fixed a potential race condition when authenticating leading to authentication failing silently ## v2.0.1 ### User Experience * Added "Copy sign in link" option on the sign-in waiting screen for cases where the browser doesn't open automatically * Landing page navbar now shows an **Account** button in place of "Try Demo" for signed-in users, linking to account settings ### Bug Fixes & Performance * Fixed "Analyze samples" context menu not queuing analysis jobs * Fixed sync pulling from remote after a failed push, which could overwrite local data * Fixed potential database corruption from background and manual syncs running simultaneously * Fixed fresh installs with sync credentials failing on a nonexistent remote database * Optimized database size and sync throughput when it's idle Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## The Rebuild: Offline-First, Faster Everything Sample Vault 2.0 is a ground-up rebuild of how the app stores and processes your data. Every operation is faster, everything works offline, and a new processing pipeline gives you more control over how your samples are organized. ### Offline-First Architecture The local database has been completely replaced. The previous in-memory engine was fragile and kept all data in RAM — the new one writes to disk, syncs reliably, and uses a fraction of the memory. * **Faster everything**: Browsing, searching, and filtering are all significantly faster * **Lower memory usage**: Data is no longer held entirely in RAM * **More stable**: Fewer crashes and edge cases from the previous in-memory database * **Full data sync**: All data in the app now syncs to the cloud — previously only a subset was synced The [AI Assistant](/core-features/ai-assistant) still requires an internet connection. All other features work fully offline after initial sign-in. ### Sound Profile A new **Sound Profile** tab in the [Sample Browser](/core-features/sample-browser) filter sidebar lets you filter samples by their sonic characteristics using range sliders. * **Core**: Tonality, Percussiveness, Brightness, Noisiness, Complexity, Attack, Rhythmic Density, Pitch Stability * **Frequency Content**: Sub Bass, Bass, Low Mids, Mids, Highs * **Stereo & Temporal**: Stereo Width, Temporal Centroid, Dynamic Range Set a min-max range on any characteristic to narrow your results. Active filters appear as removable chips in the filter rail. Sound Profile tab showing range sliders for sonic characteristics ### Sample Enrichment Sample processing is now split into two distinct stages: 1. **Analysis** — Local audio processing that extracts tempo, key, spectral features, and sound characteristics. Deterministic, fast, and free 2. **Indexing** — AI scans your directories and generates relevant labels: genres, tags, creators, and sample packs. Costs 20 credits per indexed folder — a fraction of the previous per-sample cost, which has been removed entirely After indexing, **enrichment** automatically matches the AI-generated labels to individual samples — no additional AI cost. Right-click any directory in the [File Browser](/core-features/file-browser) and select **Index Directory** or **Enrich Samples** to get started. Manage AI-detected labels in **Settings > Processing > Enrichment Labels** — control which genres, creators, sample packs, and tags are allowed to be assigned to your samples. Enrichment Labels settings showing genre, tag, creator, and sample pack controls ### Prompt Search for Everyone [Prompt Search](/core-features/prompt-search) is now available to **all users**, not just premium subscribers. Each search costs 5 credits. Free plan users receive 100 daily credits — enough for about 20 searches per day. Samples must be enriched and embedded to become available for Prompt Search. When credits are unavailable, Prompt Search falls back to a local embedding model. Results are still useful but less accurate than the AI-expanded search. ### Processing Settings New **Settings > Processing** tab gives you control over how the app uses your hardware during audio analysis. * **Processing Threads**: Control how many CPU threads are used for audio analysis — higher values process faster but use more CPU * **Classification Models**: Control how many audio classification models run simultaneously — each instance uses approximately 500 MB of RAM Processing settings showing thread count and classification model controls ### User Experience * **Copy with effects rendered**: When time stretch or key transposition is active, right-click any sample to copy it with the effect baked in using "Copy timestretched", "Copy pitched", or "Copy timestretched & pitched" * **Master volume in title bar**: A volume slider is now accessible directly from the title bar with a click-to-mute toggle * **Activity log refresh**: The activity log has been streamlined to surface only the most relevant background events, reducing clutter ### Performance and Bug Fixes * Sample analysis is up to **\~10x faster** * RAM is now properly released after analysis completes * File browser is drastically faster for large libraries (100k+ samples) * Fixed a race condition where the AI Assistant's tools would occasionally stop responding * Fixed auth deep link occasionally failing to complete sign-in **macOS users:** Please redownload the app from the website and select the correct version for your CPU architecture. A previous bug caused Apple Silicon users to receive Intel builds. **This is a breaking update.** The new database architecture is not compatible with previous versions. You will be prompted to migrate your data after updating, although it's recommended to start fresh and reanalyze your samples. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Native Experience Update The Sample Vault desktop app now runs entirely from your machine. The interface is bundled directly into the application instead of loading from a remote server. * **Instant startup**: The app launches immediately — no waiting for a remote connection or loading screen * **Full offline UI**: Every part of the interface works without an internet connection, not just the local features. Previously, a network hiccup could leave you staring at a blank screen mid-session * **Faster across the board**: Navigation, panel switching, and UI interactions are snappier since nothing is fetched over the network * **More reliable updates**: App versions are fully self-contained — what you download is what you run, with no risk of UI/backend version mismatches Cloud features like the [AI Assistant](/core-features/ai-assistant), [Prompt Search](/core-features/prompt-search), and sample sync still require an internet connection. ### Bug Fixes & Performance * Fixed a corrupt audio file being retried forever instead of failing after a few attempts, which could clog processing and slow the rest of your library Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Analysis Overhaul and Playback Controls Update ### Auto-Loop and Play Controls New **Auto-Loop** and **Play** buttons are now accessible directly from the title bar — start playback and toggle looping without reaching for a panel. ### Sample and Directory Reanalysis Right-click any sample or directory in the [File Browser](/core-features/file-browser) to trigger **reanalysis**, updating audio features without a full library rescan. ### User Experience * **Richer progress status**: Scanning and analysis now display extended progress details so you always know what's happening * **App data directory links**: **Settings → App Data** now includes direct links to reveal cache, logs, and other app-related directories on your system * **Improved playback behavior**: Pressing Space now always restarts playback from the beginning. Use Shift+Space to resume from where you left off * **New macOS icon**: Redesigned app icon that feels more native on macOS ### Performance and Bug Fixes * **10–20x faster sample analysis** — major optimizations to the audio analysis pipeline dramatically reduce processing time across your library * Fixed samples not obeying [Link](/core-features/link-tempo-key) and not playing in sync * Fixed [AI Assistant](/core-features/ai-assistant) chat input not rendering properly in certain cases * Fixed assistant sample player using outdated sample info/edit dialogs * Fixed download link on the web detecting the wrong architecture on macOS **macOS Apple Silicon users:** If you're on an M-series Mac, please download and reinstall the app from the [downloads page](https://samplevault.ai/downloads) to ensure you're running the correct native build. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Recall, Refinement, and Waveform Update ### Recall Tab Sample Vault now remembers what you do. The new **Recall** tab keeps a running history of every sample you drag, copy, search for, or ask the assistant about — organized by workspace. * Tracks five interaction types: samples dragged to DAW, samples copied, browser searches, [prompt searches](/core-features/prompt-search), and assistant messages * Filter by interaction type to quickly find what you were working with * Embedded sample players let you re-audition directly from the history * **AI-aware context**: Your recent activity is automatically passed to the [AI Assistant](/core-features/ai-assistant), so it understands what you've been working with and gives more relevant responses — no need to re-explain Recall tab with interaction history and embedded sample players Refer to the [full feature documentation](/core-features/recall) for more details. Recall works best when you work within specific workspaces. Workspace-scoped history keeps context focused, so the AI Assistant gives more relevant responses without activity from other projects bleeding in. ### Similar Samples Refinement The [Similar Samples](/core-features/similar-samples) panel now lets you **approve or reject** results to steer the similarity search toward exactly what you need. * Hover over any result to reveal a **checkmark** (more like this) and **X** (not like this) * Approved samples pull the search toward their sonic characteristics; rejected samples push it away * Results update in real time as you refine * Click **Reset** to return to the neutral similarity search Under the hood, approve/reject works by adjusting the search vector using the embeddings of your marked samples — blending toward approved and away from rejected — so each refinement meaningfully reshapes the results. ### Waveform Visualizer Tab A new **Waveform Visualizer** panel provides a detailed, interactive view of the currently playing sample. * Click anywhere on the waveform to seek playback * Scroll to zoom in/out (cursor-anchored) * Shift+scroll (or trackpad horizontal scroll) to pan * Enable **[Tempo Sync](/core-features/link-tempo-key)** to display beat and bar markers aligned to the sample's BPM — musical intervals replace the default time ruler The visualizer uses a multi-LOD WebGL2 rendering pipeline with sinc interpolation at high zoom levels, so it stays smooth whether you're viewing the full waveform or zoomed into individual samples. ### Improved Tempo and Loop Detection The audio analysis pipeline now uses a **multi-algorithm approach** for more accurate tempo and loop detection. * Tempo detection combines spectral flux, high-frequency content analysis, phase deviation, and inter-onset interval histograms to produce more reliable BPM estimates * Loop detection classifies samples as loops or one-shots using separate heuristics for percussive, tonal, and mixed content * Grid-based matching cross-references detected BPM against expected loop durations (1, 2, 4, 8, 16 bars) for higher confidence on loops ### Also New * **Sample Info player**: The Sample Info tab now includes an embedded audio player with waveform visualization at the top — audition and inspect without leaving the detail view * **UI scale setting**: Adjust the interface scale in **Settings → Appearance** to match your display and preferences * **Web browser demo**: Try Sample Vault directly in your browser at [samplevault.ai/demo](https://samplevault.ai/demo) — no download required * **Public roadmap**: See what's coming next and vote on what's important to you on the [public roadmap](https://samplevault.ai/roadmap) ### User Experience * **Sample reconciliation**: If samples are moved or deleted from disk, Sample Vault detects the missing file on playback and offers to **locate** the file at its new path or **remove** it from your library. Relocating a sample automatically finds and updates other samples from the same original directory. * Extended activity log with more processing events * [AI Assistant](/core-features/ai-assistant) now has access to the most up-to-date documentation * New progress bar in the [Sample Browser](/core-features/sample-browser) shows analysis progress more cleanly — samples now auto-refresh every 60 seconds instead of continuously, reducing browsing interruptions * BPM sync now falls back to 120 BPM when no tempo is detected, instead of showing an empty state * **(Experimental)** Database backups available in **Settings → App Data** — may not handle very large datasets yet * Right-clicking the user icon now shows the same menu as left-clicking * Added a default right-click context menu anywhere in the app with links to keyboard shortcuts, settings, Discord, documentation, changelog, and more ### Performance and Bug Fixes * Audio analysis processing speed improved up to \~5× * Reduced RAM usage during processing * Improved [Prompt Search](/core-features/prompt-search) accuracy * Fixed Assistant occasionally responding with empty messages due to invalid tool calls * Fixed playing assistant samples not triggering the [Similar Samples](/core-features/similar-samples) panel to update * Fixed chat context menu action redirecting to an old route * Fixed samples not being properly deleted during server sync * Fixed samples not grouped correctly into original directories when pulling from server sync * Fixed errors when extracting metadata from files during processing * Fixed virtual file tree being built incorrectly ([Settings](/reference/settings) → App Data → Rebuild File Tree if needed) * Fixed system status button in the title bar continuing to play sheen animation after analysis finishes * Fixed samples being visible across accounts on the same PC * Fixed links in [AI Assistant](/core-features/ai-assistant) chat not opening in the browser window Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Dock Layouts Update ### Dock Layouts Sample Vault now has a fully customizable layout engine. Arrange any combination of panels — sample browser, file browser, AI assistant, similar samples, note editor, and more — into your own tabbed workspace. * Drag and drop panels to rearrange or split the interface * Resize panels to fit your workflow * Save and switch between named layouts * Use as many or as few panels as you need Refer to the [full feature documentation](/core-features/dock-layouts) for more details. ### Improved Sample Info Tab The sample info tab has been redesigned with three focused tabs: **General**, **Characteristics**, and **Metadata**. * Characteristics tab shows all audio features extracted during processing (tempo, key, spectral data, and more) * Editing sample metadata is faster with a cleaner layout * Tab automatically refreshes based on the last selected sample ### Improved File Browser Tab The file browser now more closely mirrors how DAWs handle file trees. * Collapse and expand directories just like in your DAW * Right-click a directory to mark all samples underneath for upload/analysis, or delete them by directory * Collapse All button for quickly folding the entire tree * Search bar to filter by filename * Virtualized rendering for smooth scrolling across very large libraries * Add new samples directly from the browser ### New Similar Samples Tab A new **Similar Samples** tab automatically surfaces the most sonically similar samples whenever you play one. No searching required — just audition a sample and the tab updates instantly. ### Control Which Samples Get Analyzed and Uploaded You can now choose which samples get analyzed and synced to the cloud — both during the initial directory import and at any point afterward. * Configure during the **Add Sample Directory** flow * Right-click individual samples or entire directories in the File Browser to update the setting at any time Samples that were already analyzed or uploaded before this update cannot be retroactively changed to local-only. To keep them local, delete them from your library and re-add the directory with analysis and upload disabled. ### Activity Log All background events are now captured in an **Activity Log**, viewable in **Settings → Activity Log**. See a full timeline of sample processing steps and cloud sync events in one place. ### New Note Editor The note editor has been rebuilt with richer authoring support. * Full Markdown support — [see Markdown reference](https://www.markdownguide.org/basic-syntax/) * Slash commands for quick formatting * Drag and drop to reorder sections * Create to-do lists with `- [ ]` syntax — previously separate to-do lists are now consolidated into notes ### User Experience * **Discord OAuth** is now available as a sign-in option * **Workspace selector** has moved to the title bar as a persistent global switch (previously called Projects) * When switching workspaces, the key and BPM are automatically applied to playback * Sample processing and sync status are now consolidated into a single **System Status** indicator with more detail and recent activity log entries System status indicator showing processing and sync state ### Performance and Bug Fixes * Drastically improved sample player rendering performance * Prompt Search results are now more accurate * Prompt Search results now display a similarity level next to each result * Fixed an error that occurred when using "Reveal in Path" on certain samples * Changed the icon to look better on MacOS * Fixed MacOS tray icon not showing menu actions on click * Fixed tray icon not having proper tooltip Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Quick Actions Update ### Quick Actions Press `Cmd/Ctrl + Shift + Space` to open the new Quick Actions menu — a fast way to search and filter your sample library using search phrases and tokens. Tokens let you narrow results instantly: * `!loop` / `!oneshot` — filter by sample type * `@Cmaj` / `@F#` — filter by key * `#120` / `#140-160` — filter by tempo or tempo range * `$edm` — filter by genre * `%dark` — filter by mood Quick Actions also leverages Link, key/pitch lock, and tempo lock to synchronize playback directly from search results. Refer to the [full feature documentation](/core-features/quick-actions) for more details. ### Sample Browser Filters Some of the sample browser filters have been rewritten. The were two goals with these changes: * Make sample filtering more efficient by eliminating redundant interactions * Provide a visual representation of possible options without having to click and browse through menus New sample browser filters panel ### User Experience * Restored the ability to delete sample directories * Key transposition buttons no longer jump unexpectedly * Expanded the user profile menu with additional links * AI Assistant chat now properly displays the character limit in the prompt instead of silently crashing * Added feedback buttons below assistant messages to improve model behavior over time New feedback buttons for assistant messages ### Performance and Bug Fixes * Fixed a critical bug on Windows where undownloaded samples would fail to download/play, showing an unexpected error * Fixed samples occasionally not playing at random * Fixed waveform progress not displaying correctly * Improved synchronization timing with Link * Improved sample categorization into virtual directories — fewer misassigned samples (e.g., samples incorrectly tagged as hats) * Time-stretching and key transposition are now bypassed when not in use, preserving original audio quality * Optimized rendering in the Sample Browser and other views with sample players * Uploads now handle choppy network conditions more gracefully and wait to reconnect before retrying Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ## Link, Tempo and Key Lock Update ### Lock Playback Tempo and Key Sample Vault now has a time-stretching and key transposing algorithm built-in. Globally define the tempo and key at which all samples play. Tempo lock requires samples with detected tempo. Key transposition requires samples with a detected key. ### Link — Synchronize Playback With Your DAW Sample Vault now supports [Ableton Link](https://www.ableton.com/en/link/) to synchronize sample playback with your DAW. Enable it by clicking the **Link** button in the title bar. See your DAW's documentation for how to enable Ableton Link — most major DAWs support it, including [Ableton Live](https://www.ableton.com/en/manual/synchronizing-with-link-tempo-follower-and-midi/), [Bitwig](https://www.bitwig.com/userguide/latest/the_dashboard/), and [Logic Pro](https://support.apple.com/guide/logicpro/ableton-link-in-logic-pro-for-mac-lgcp528602e4/mac). When Link is enabled, sample playback loops until paused. Enabling Link automatically locks tempo playback. Refer to the [full feature documentation](/core-features/link-tempo-key) for more details. ### Sample Processing * Sample directory view now shows richer details per sample and per processing step * Samples that fail to process or upload now automatically retry * More descriptive error messages for common processing errors * Directory scans now automatically resume if interrupted Sample directory view showing detailed sample metadata and processing status ### User Experience * New waitlist on the [pricing page](https://samplevault.ai/pricing) for a one-time purchase license * Removed upsell banner from sample browser for free plan users * New onboarding survey for a more personalized experience * Smoother application update flow for future updates * Public sample libraries are bundled with the app for faster first launch ### Performance and Bug Fixes * Sample browser search is significantly faster * Fixed AI Assistant chat errors caused by invalid tool calls * App no longer crashes after server-side updates — it gracefully reloads instead Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). ### Known Issues * **Link peer detection:** Link may occasionally fail to detect other peers on the network. Toggle Link off and on to reconnect. * **Drag-and-drop with lock enabled:** Dragged samples do not yet preserve tempo/key lock settings. This will be addressed in a future update. This update is a breaking change. You must install the newest version of Sample Vault to use the application. You will be prompted to do so on startup if you haven't done already. You can manually download and update the app from the [downloads page](https://samplevault.ai/downloads). ## Offline Mode and Unlimited Free Plan Update ### Offline Mode (Local-First) Sample Vault now works fully offline. * **Zero Latency:** Searching and organizing happens locally on your device * **Auto-Sync:** Changes made offline automatically sync to the cloud once you reconnect (paid plans only) The AI Assistant and Prompt Search still require an internet connection. You must sign in while online before using offline mode. ### Unlimited Free Plan The Free Plan no longer has sample limits. Add as many samples as you want — all samples are analyzed locally. Cloud sync and AI analysis remain exclusive to [paid plans](/reference/plans-usage). ### AI Assistant: Fast Mode New **Fast Mode** toggle for the AI Assistant, optimized for quick questions, precise sample searches, and documentation lookup. Dropdown menu showing the new AI Assistant Fast Mode option ### Sample Libraries Management Enable or disable specific public sample collections to declutter your library. Configure this in **Settings → Sample Libraries**. Public sample libraries management interface ### Performance and Bug Fixes * **Search performance:** Sample browser loads faster and scrolls more smoothly * **AI Assistant reliability:** Reduced hallucinations and significantly lower error rate No breaking changes in this release. # AI Assistant Source: https://docs.samplevault.ai/core-features/ai-assistant Chat-based sample discovery and music production help ## Overview The AI Assistant is a chat panel that searches your sample library, suggests samples by description, and answers production questions. It can draw on every sample in your processed library, on the manuals for major DAWs and plugins, and on Sample Vault's own documentation — so "how do I sidechain this?" and "how does indexing work?" are both fair game. Answers pulled from documentation link back to the page they came from. ## Modes The dropdown next to the chat input switches between two response modes: * **Creative** (default) — more expressive, exploratory responses. Best when you want variation, fresh ideas, or arrangement and composition suggestions. * **Fast** — quick, lighter responses for simple requests, documentation questions, and general chat. On **Demo Room** — and a Lifetime license without a subscription — the assistant draws from a free weekly **AI actions** allowance shared across the assistant's AI features. **Home Studio** includes a monthly assistant allowance; **Professional Suite** runs the assistant under fair use with no monthly cap. See [Plans & Usage](/reference/plans-usage#starter-pools). ## Chat input The chat input supports multi-line text and submits on Enter. The button at the far left of the input opens your chat history. The lightbulb button to the right offers categorized prompt suggestions you can insert and edit before sending. ## Sample results When the AI finds samples, they appear inline as collapsible playable cards. Each card includes: * Play/pause and waveform * **Copy** to clipboard * **Drag handle** — drag straight into your DAW * **Info** for full metadata * **Right-click menu** — find similar, get creative suggestions, analyze, add to favorites or collections, show other samples in the same directory The AI returns one sample per directory by default to keep results varied. Use **Show other samples in directory** from the right-click menu to see the rest of a folder. When a reply includes more than one sample, an actions row beneath it works on the whole set at once: copy them all to your clipboard (applying your playback cleanup the same way copying or dragging a single sample does), add them all to a collection, send them to [Sketch](/core-features/sketch) as a kit, or select them all to drag as a batch or bulk-edit. ### Selecting chips across the chat `Cmd/Ctrl + click` a sample chip in any message to add it to the selection. `Shift + click` ranges within a single message. With a chip focused (or an existing selection active), `Cmd/Ctrl + A` selects every chip across the whole chat — when nothing is focused, it falls through to text selection as you'd expect. Selected chips feed the same selection bar used everywhere else, so you can drag the whole batch into your DAW or bulk-edit through the **More** menu. See [Multi-select & batch actions](/core-features/sample-browser#multi-select-batch-actions). ## Building a kit Ask the assistant for a kit or a starting beat and it assembles one from your own library — choosing sounds that fit the vibe you describe and saving them as a named kit you can reopen later. The reply includes a link that opens the kit in [Sketch](/core-features/sketch); when Sketch is already open with unsaved changes, it confirms before replacing them. Nothing is synthesised — every pad is a file you already own. ## Building a collection Describe a slice of your library and the assistant gathers it into a named [collection](/core-features/sample-browser), then opens the Library scoped to it so you see the result straight away. You can also ask it to show any collection you already have, and it narrows the Library to that one. ## Projects A [project](/core-features/projects) carries context — BPM, key, genre, mood, notes, todos — that the AI reads on every message. Switch between projects from the title bar [Project Selector](/reference/ui-reference#title-bar-controls). Project notes live in the [Note Editor](/core-features/note-editor); recent activity feeds the AI through [Recall](/core-features/recall). When a [DAW session](/core-features/projects#daw-projects) is open, the assistant can also look inside the project file on request — the tracks, the samples and plugins in use, the tempo and key — and cross-reference those samples against your library to say which you already own and what it knows about them. ## Memory The assistant keeps track of what matters between sessions, so you stop re-explaining yourself each time you open a chat. It remembers two kinds of things: * **Durable facts** — your preferences, the conventions you work by, and things about your library, like the genres you own and your go-to tempo ranges. These stay until they change. * **Current work** — lighter context about what you're in the middle of. This fades on its own if it stops coming up, so a one-off session doesn't linger. The assistant builds and refreshes this on its own from your recent conversations and library activity, about once a day. You can also tell it to remember something the moment it matters — or to forget it — and it takes effect right away. Everything the assistant remembers is visible and editable in [Settings](/reference/settings#assistant-memory), alongside your assistant instructions: read any entry, edit or delete it, switch single entries off, or turn memory off entirely. When an entry is flagged as possibly out of date, you decide whether to confirm or dismiss it. What the assistant remembers stays in your library's database on your machine. Memory is part of **Professional Suite**, or **Lifetime** with your own AI key. On **Home Studio** and **Demo Room**, the assistant doesn't carry memory between sessions. ### Measured context Separately from memory — and on every plan — the assistant reads a measured picture of your library and your current session: the genres you own versus the ones you actually reach for, the tempo range you work in, the searches you keep coming back to, and what you've searched and played this session. You don't manage this and it isn't remembered; it's recomputed from your activity, so it stays current on its own. ## Tool calls When the AI searches or fetches data, you'll see a status indicator: a spinner while running, a checkmark when complete. Common actions: "Searching for samples...", "Finding similar samples...", "Browsing local files...", "Fetching sample details...". ## Troubleshooting ### The AI gives generic responses instead of understanding my style Add more detail to your **AI Assistant Instructions** in [Settings](/reference/settings#ai-assistant-instructions). Replace "I make electronic music" with something like: "I produce melodic techno at 120–125 BPM. I prefer organic percussion over synthetic drums and synth patches with movement and modulation." # Dock Layouts Source: https://docs.samplevault.ai/core-features/dock-layouts Build a custom workspace by arranging panels the way you work ## Overview Dock Layouts let you build your own Sample Vault workspace. Arrange any combination of panels — library, file browser, AI assistant, similar samples, note editor, and more — into a tabbed layout that matches how you actually produce. ## Creating a Layout Start a new layout and you're offered arrangements built around what you're about to do — browsing your folders alongside the library, producing with the assistant and notes on hand, exploring the map, inspecting what's playing, watching it as sound, matching a reference track, sketching a pattern, or organizing what you're working on. Each one arrives with its panels already in place, ready to work in. Prefer to fill it yourself? **Customize** hands over the bare shapes instead — plain columns, rows, columns with one side split, and grids — with empty slots to populate however you like. Either way the shape you pick fixes the number of slots: swap what's in a slot, move panels between slots, and resize by dragging the dividers, but the arrangement itself holds. At the bottom of Customize there's also a fully flexible layout where dragging a panel to the edge of another carves out a new slot — the older behavior, and easy to end up somewhere you didn't intend. Layouts save automatically as you work. ## Available Panels | Panel | Description | | - | - | | Library | Browse and filter your full sample library | | File Browser | Navigate your directory tree and manage samples | | AI Assistant | Chat-based sample search and recommendations | | Similar Samples | Automatically surfaces samples similar to what you're playing | | Suggestions | Sample picks based on recent searches, drags, and your project | | Note Editor | Write notes and to-do lists for your projects | | Recall | View a timeline of your recent interactions and activity | | Waveform | Display the waveform of played samples | ## Switching Between Layouts Open the Layouts menu at any time to switch to a saved layout. Switching is non-destructive — your previous layout is always preserved. Your first nine layouts also get automatic keyboard shortcuts (Alt+1 through Alt+9, in list order); each layout's tooltip shows its shortcut. ## Modifying and Deleting Layouts * **Rename**: Open the layout menu, click the layout name, and type a new one * **Reset**: Drag panels back into your preferred arrangement and the layout saves automatically * **Delete**: Remove a layout from the layouts menu — this does not affect your samples or notes A layout built from a fixed arrangement keeps its slots for good: a slot you empty stays open and waiting rather than collapsing into its neighbour. ## Tips Create one layout for **browsing** (file browser + library + similar samples) and another for **composing** (assistant + note editor + a narrow library). Switch between them with one click as your session shifts from finding sounds to building ideas. You can open the same panel in multiple panes — for example, two Library instances with different filters applied side by side. # Duplicate Detection Source: https://docs.samplevault.ai/core-features/duplicate-detection Find and collapse copies of the same sound across your library ## How It Works When a sample is analysed, Sample Vault gives it a **signature based on what it actually sounds like** — not its filename, size, or tags. Samples whose signatures match are grouped into a duplicate cluster. Because the signature describes the audio itself, it catches copies a filename or file-size check never could: * The same sound saved as **WAV, AIFF, or FLAC** * Exported at **16- or 24-bit**, or at **44.1 or 48 kHz** * **Normalised** or otherwise gain-adjusted * Re-saved with **different metadata** or a different filename One copy of each cluster — the **kept copy** — is the one that surfaces when you search, filter, or look for similar sounds. The rest stay out of those results. Hidden copies are never deleted; your files on disk are left exactly as they are. Browsing by folder is the exception: a folder always lists everything that's actually in it, with hidden copies marked as duplicates. That way a folder's contents match its file count, and you can see at a glance which folders are redundant before deciding what to do about them. Duplicate detection works on samples that have been analysed. Older samples join in as their directories are re-scanned. ## Which Copy Is Kept Sample Vault picks the best copy of each cluster to keep visible. A copy you've **favorited** or added to a **collection** always wins — Sample Vault never hides something you've deliberately organised. Otherwise it prefers the higher-quality file: lossless over lossy, higher bit depth and sample rate, real metadata over "Unknown", and a tidier file path. ## Settings › Duplicates The **Settings → Duplicates** tab shows what's been collapsed and puts you in control: * **Collapse duplicates** — the master switch. Turn it off to see every copy in search and filter results as well. * **Match strictness** — how alike two samples must be to count as duplicates: * **Exact** — the same decoded audio (the default). * **Same sound** — also catches lossy re-encodes and tiny edits. * **Near-identical** / **Looser** — progressively fuzzier. The loosest setting can group sounds that are merely very similar, so review the results before relying on it. * **Scan now** — re-check the library on demand. A scan also runs automatically after every directory scan. * **Review** — browse every duplicate group, flip individual copies between *kept* and *hidden*, and collapse groups as you clear them. * **Move to Trash** — reclaim disk space by sending the hidden copies to your system Trash. They stay recoverable there until you empty it; the kept copy is never touched. You can watch it run and stop it partway, and anything your system refuses to move keeps its place in the library so you can see what was left behind. Looking for samples that are *similar* rather than identical? That's what [Similar Samples](/core-features/similar-samples) is for — duplicate detection is about copies of the *same* sound. # File Browser Source: https://docs.samplevault.ai/core-features/file-browser Navigate, manage, and control your sample directories ## Overview The File Browser gives you a DAW-style view of your sample directories. Navigate your folder structure, manage which samples get analyzed and uploaded, and control your library directly from the file tree. ## Navigating the Tree Directories collapse and expand just like in your DAW. Click any folder to expand it and reveal its contents, click again to collapse it. * **Search bar**: Filter the file tree by filename to quickly locate a specific sample or folder * **Collapse All**: Fold the entire tree in one click to get back to a clean top-level view The file browser uses virtualized rendering, so it stays fast and smooth even with tens of thousands of samples across deeply nested directories. ## Adding Samples Click the **Add Samples** button at the top of the file browser to add a new directory. You'll be prompted to choose a folder and configure analysis and upload settings before the scan begins. See [Adding Samples](/get-started/adding-samples) for a full walkthrough. ## Right-Click Actions Right-clicking gives you contextual actions for both directories and individual samples. ### On a Directory * **Key** / **Scale** / **Type** — set the value on every sample under this directory (confirms before applying) * **Tempo** — pick a common tempo or type a custom BPM; applied to every sample under this directory * **Creator** / **Sample pack** — search existing values or type a new one; applied to every sample under this directory * **Category** — assign a category to every sample under this directory * **Tags → Add tags / Remove tags** — stage tags and apply to all samples. Remove shows only tags that already exist in this directory * **Genres → Add genres / Remove genres** — same but for genres * **Analyze samples** — queue every sample in this directory (and subdirectories) for audio processing * **Classify samples** — re-run AI audio tagging on every sample under this directory without redoing local analysis * **Upload samples to the cloud** — sync samples to cloud storage for backup and cross-device access * **Enrich samples** — match AI-generated labels from Deep Tagging to individual samples in the directory * **Deep-tag directory** — run Deep Tagging to generate genre, tag, creator, and sample pack labels (whole library on paid plans; one free folder on Demo Room). See [Deep Tagging](/core-features/sample-enrichment) * **Delete all samples** — remove all samples in this directory from your library ### On a Sample * **Analyze sample** — process this sample individually * **Classify sample** — re-run AI audio tagging without redoing local analysis * **Upload to cloud** — sync this sample to cloud storage * All standard [sample context menu](/reference/ui-reference#sample-context-menu) actions are also available ## Multi-select `Cmd/Ctrl + click` toggles, `Shift + click` ranges across the visible tree, and `Cmd/Ctrl + A` grabs every sample in the expanded folders. Selection feeds the same bar — and the same batch actions — used everywhere else. See [Multi-select & batch actions](/core-features/sample-browser#multi-select-batch-actions) in the Library docs. ## Controlling Analysis and Deep Tagging When you add a new directory, you choose whether its samples get analyzed automatically. You can also deep-tag a directory from its right-click menu. See [Adding & Managing Samples](/get-started/adding-samples) for what processing unlocks. # Filters Source: https://docs.samplevault.ai/core-features/filters Narrow your results from the search bar, the Filters dialog, or saved presets ## Overview Two surfaces drive what you see in the [Library](/core-features/sample-browser) results: the **search bar** (your master control) and the **Filters** dialog beside it, for browsing values you don't already know by name. Both stay in sync. Picking a tag in the dialog inserts a `%tag` chip in the search bar; typing `@Cmin` selects that key. For filtering by sonic characteristics like tonality, brightness, or stereo width, use the [Sound Profile](/core-features/sound-profile) section of the dialog — or type characteristic tokens directly (`atk:30-80`, `bright:0-50`). ## Search bar (master control) Every filter can be expressed as a typed token. See [Search Syntax](/core-features/search-syntax) for the full grammar — short prefixes like `%tag`, `$genre`, `&creator`, `+pack`, `*` (favorites), and long forms like `category:`, `collection:`, `sort:`, plus all 16 sound profile characteristics. Selected filters render as colored chips inside the input. **Click any chip to remove it.** The **×** button on the right clears every filter at once. The smart suggestion dropdown helps you discover values without typing the syntax: type two characters of a name and pick from the grouped list of matching tags / genres / creators / packs / collections. The same dropdown also surfaces field shortcuts (`creator:`, `pack:`, etc.) so you can learn the syntax by using it. ## Filters dialog The **Filters** button next to the search bar opens every filter in one dialog, alongside sort order, how results are displayed, and your saved filters. Use it when you don't already know which tag, creator, or pack you want — each section lists every value with sample counts, and its own search field jumps straight to a value anywhere in the dialog. Changes apply to the results as you make them, so you can leave it open while you narrow down. Sections: * **Category** — filter by directory or virtual directory, with drill-in chips * **BPM range** — minimum and maximum tempo * **Key & Scale** — musical key (C, D#, F, etc.) and scale (major, minor, modal) * **Collections** — jump to a saved collection, or create a new one (AI Collections available on paid plans) * **Sample Pack** and **Creator** — filter by detected pack or artist * **Type** — loops, one-shots, or all * **Sort** — by relevance, date added, or name, plus by sample length, tempo, loudness, or any sound-profile characteristic (each ascending or descending) * **Genres** — AI-detected or manually assigned * **Tags** — your custom tags plus auto-generated ones Genres and tags are organized into families rather than one flat list, so the picker stays short even on a large library. Pick a family to match every value under it at once, or open it to choose a single genre or tag. Sample Vault keeps these families tidy on its own as your library grows — see [Deep Tagging](/core-features/sample-enrichment) for what that involves. Filters can also **exclude**: alongside picking a tag, genre, or category to keep, you can filter one *out*, dropping every sample that carries it. Exclusions show as their own `not:` chips and clear the same way as any other filter. [Deep Search](/core-features/search-modes) uses them too — ask for "no vocals" and the exclusion lands as a visible chip you can remove. Whatever you've applied shows as chips under the search bar, so the results keep full width and the row stays empty until something is actually filtering. Clearing everything and saving the current set as a preset both sit beside the **Filters** button; the shortcuts are listed under [Keyboard Shortcuts](/reference/keyboard-shortcuts). **Rename or delete library-wide**: right-click any tag or genre chip — in the Filters dialog or in the Sample Info pane — and choose **Rename** or **Delete**. Rename updates the value on every sample that has it; Delete strips it from every sample after a confirmation. ## Filter presets Save a filter combination once you've got it right, and reload it later. * Save the active state under a name from the control beside the **Filters** button * Restore a saved set from **Saved filters** in the dialog * Delete one you no longer want from the same place Presets capture every active filter — tokens in the search bar, values picked in the dialog, sort order, and sound profile positions. # Circle of Fifths Source: https://docs.samplevault.ai/core-features/library-circle Browse your library by musical key on the classical circle of fifths ## Overview The **Circle** view in the Library lays every sample out by its musical key, on the classical circle of fifths. Each sample is its own dot, the same way it is on the Map — the difference is *where* a sample lands. On the Map dots cluster by sonic similarity. On the Circle they cluster by tonality. It's a quick answer to "what do I have in F minor at 138 BPM?" — and a way to scan whole regions of your library you may have forgotten about. ## Switching to Circle Pick **Circle** from the Library's display options. Each Library panel remembers its own view mode independently, so you can keep one panel on the list and another on the Circle if you split your dock. ## Reading the wheel * **Outer ring — twelve majors.** C/Am at 12 o'clock, going clockwise by fifths through G, D, A, E, B, F♯, C♯, G♯, D♯, A♯, F. * **Inner ring — twelve relative minors.** Each major's relative minor sits directly inside it — A minor inside C major, E minor inside G major, B minor inside D major, and so on. This is the classical pairing; samples that share a key signature share a wedge. * **Centre disc — keyless and modal samples.** FX, percussion, drum hits without a detected pitch, and samples tagged with a modal scale (Dorian, Lydian, Phrygian, etc.) land here. A small label under the wheel shows how many. ## Radial axis The radial dropdown in the Circle header decides what the *distance from centre* means inside each wedge: low values sit near the inner edge, high values near the outer edge. You can choose any of the sample's sound characteristics — tempo, brightness, attack, stereo width, loudness, duration, and the rest of the [Sound Profile](/core-features/sound-profile). Flipping the axis rearranges dots radially without moving them angularly — the C-major wedge stays C-major, the dots inside it just spread differently. Use this to scan, for example, "fast bright C-minor stabs" by setting the axis to brightness and looking at the outer ring of the C/Am wedge. ## Interactions Everything from the Map works here too. * **Click a sample** to play it through the audio player. * **Drag a sample** straight into Ableton, Logic, FL Studio, or any DAW that accepts file drops. * **Right-click a sample** for the full sample menu — favourite, add to collection, set key/scale/tempo, classify, upload, copy, delete. * **Shift-drag** to lasso a region and multi-select. Save the result as a collection or apply it as a filter on the List view. * **Arrow keys** step through the visible samples in the direction you press, just like on the Map. * **Mouse wheel** zooms; drag empty space to pan, or hold the middle mouse button to pan from anywhere. And one gesture unique to the Circle: * **Click a key label** (e.g. the `F♯` text on the outer ring, or `Dm` on the inner ring) to filter the library to that key + scale. The key and scale filters update in lockstep — the Circle and the filters are bidirectional. ## How the Circle relates to filters Active filters dim non-matching samples the same way they do on the Map. If you've filtered to `loop` + `genre: techno`, the wedges still show the wheel geometry but only matching dots stay at full brightness; the rest fade. **Isolate** in the Circle header hides the dimmed dots entirely. The chip filter for key + scale is bidirectional: click a wedge label, and the chip lights up; pick a key in the chip filter, and the matching wedge stays bright while the rest of the wheel dims. ## Searching on the wheel Describe a sound in the search bar — or run [Deep Search](/core-features/search-modes) — and the Circle becomes a map of your matches — the closest samples stay at full brightness while everything else dims, so you can see *where* on the wheel your matches sit. **Isolate** hides the rest entirely, leaving only the results in their key positions. Suggestion chips appear in a strip above the wheel. ## Zooming and clusters When you zoom out far enough that dots would overlap, the Circle thins each crowded area down to a few representatives. Those surviving dots grow to show they stand in for a cluster; as you zoom back in, they shrink and split into the individual samples behind them. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). # Sample Map Source: https://docs.samplevault.ai/core-features/library-map Lay your whole library out as a field where samples that sound alike sit close together ## Overview The **Map** view in the Library lays your samples out as a field of dots, with samples that *sound alike* placed close together. Walk across the field and the character shifts under you — one neighbourhood is airy pads, another is punchy kicks, another is gritty noise — without you ever typing a word. It's the view to reach for when you don't have a search in mind: a way to rediscover the corners of a large library you'd forgotten you owned, and to see the overall shape of what you've collected. The Map is also the parent of the other spatial views — [Circle of Fifths](/core-features/library-circle) and [Scatter Plot](/core-features/library-scatter) lay the same dots out by key and by two chosen characteristics; the interactions below carry across all three. ## Switching to Map Pick **Map** from the view switcher in the Library toolbar. Each Library panel remembers its own view independently, so you can keep one panel on the list and another on the Map if you split your dock. ## Reading the field * **Position means similarity.** Two dots near each other sound similar; distant dots don't. Tight clumps are pockets of closely related samples. * **Colour means category.** Each dot is tinted by its top-level category, and a legend names the colours, so you can see at a glance where your drums sit versus your pads versus your FX. The layout is built from the same sound analysis that powers [Deep Search](/core-features/search-modes) — it's reading what each sample sounds like, not its filename, so similar sounds land together even when their names have nothing in common. ## Scope The scope control focuses the Map on either your whole library or a single top-level category. Narrowing to one category re-lays just that slice across the whole field, so distinctions that were squeezed together in the global view spread out — useful for telling apart hundreds of kicks, or every pad you own, in detail. ## Building the Map The Map needs a minimum number of processed samples before it can lay anything out. Until you've added enough, it shows how many more it's waiting on — keep adding directories and it fills in. After big changes to your library you can **Regenerate** the layout from the view's overflow menu to fold the new samples in. ## Interactions These gestures work on the Map, and the same ones carry across the Circle and Scatter views. * **Click a dot** to play that sample. * **Drag a dot** straight into Ableton, Logic, FL Studio, or any DAW that accepts file drops. * **Right-click a dot** for the full sample menu — the same favourites, collections, edits, and batch actions you get in the [list view](/core-features/sample-browser). * **Shift-drag** to lasso a region and multi-select everything inside it. Save the result as a collection, or apply it as a filter back on the list. * **Arrow keys** jump to the nearest sample in the direction you press and play it, recentring the view if the sample would fall off-screen. * **Mouse wheel** zooms; drag empty space to pan, or hold the middle mouse button to pan from anywhere, including from on top of a sample. **Fit** frames your whole library again in one click. ## Filters and Deep Search on the Map Active filters dim the dots that don't match instead of removing them, so the field keeps its shape while your matches stay bright. **Isolate** hides the dimmed dots entirely, leaving only the matching set in place. Search behaves the same way: type a description in the search bar — or run [Deep Search](/core-features/search-modes) for a more precise AI pass — and the Map becomes a picture of your matches, the closest samples staying bright while the rest fade. Because position still means similarity, you can see *where* your matches sit and what else is nearby. ## Clusters When you zoom out far enough that dots would pile on top of each other, the Map thins each crowded area down to a few representatives so the field stays readable. Those stand-in dots grow to show they cover a cluster; zoom back in and they shrink and split into the individual samples behind them. Turn the behaviour off with the **Clusters** toggle in the overflow menu if you'd rather always see every dot. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). # Scatter Plot Source: https://docs.samplevault.ai/core-features/library-scatter Plot your library on two sound characteristics you choose for the X and Y axes ## Overview The **Scatter** view in the Library plots every sample on a grid, with two sound characteristics you pick driving its horizontal and vertical position. Each sample is its own dot, the same way it is on the Map and the Circle — the difference is *what* the position means. On the Map dots cluster by overall sonic similarity; on the Circle they cluster by key; on the Scatter you decide the two qualities being compared. It answers questions the other views can't frame directly — "show me the bright, snappy sounds" is the top-right corner when X is attack and Y is brightness — and it makes the trade-offs in your library visible at a glance. ## Switching to Scatter Pick **Scatter** from the Library's display options. Each Library panel remembers its own view mode independently, so you can keep one panel on the list and another on the Scatter if you split your dock. ## Choosing the axes Two dropdowns in the Scatter header set the **X** (horizontal) and **Y** (vertical) characteristics. Low values sit to the left and bottom; high values to the right and top. Pick from any of the sample's sound characteristics — tempo, brightness, attack, stereo width, loudness, duration, and the rest of the [Sound Profile](/core-features/sound-profile). * A **swap** control flips the two axes when you want to look at the same pair the other way round. * Axis labels along the bottom and left edges name the chosen characteristics and which end is which. * Changing an axis re-plots the dots in place — nothing is added or removed, the cloud just rearranges. Because both axes are sound characteristics, a corner of the plot is a precise sonic description: with X = brightness and Y = stereo width, the top-right gathers your bright, wide sounds and the bottom-left your dark, narrow ones. ## Incomplete samples A sample needs a value on both chosen characteristics to be placed exactly. Samples still waiting on analysis, or missing one of the two values, are parked near the middle of the axis they're missing so nothing silently disappears. You can hide them instead from the Scatter header's overflow menu, leaving only fully-placed samples on the grid. ## Interactions Everything from the Map and Circle works here too. * **Click a sample** to play it through the audio player. * **Drag a sample** straight into Ableton, Logic, FL Studio, or any DAW that accepts file drops. * **Right-click a sample** for the full sample menu — favourite, add to collection, set key/scale/tempo, classify, upload, copy, delete. * **Shift-drag** to lasso a region and multi-select. Save the result as a collection or apply it as a filter on the list view. * **Arrow keys** step through the visible samples in the direction you press. * **Mouse wheel** zooms; drag empty space to pan, or hold the middle mouse button to pan from anywhere. ## How the Scatter relates to filters Active filters dim non-matching samples the same way they do on the Map and Circle. If you've filtered to `loop` + `genre: techno`, the grid still shows every dot's position but only matching dots stay at full brightness; the rest fade. **Isolate** in the Scatter header hides the dimmed dots entirely, leaving just your filtered set spread across the two axes. ## Searching on the grid Describe a sound in the search bar — or run [Deep Search](/core-features/search-modes) — and the Scatter becomes a map of your matches — the closest samples stay at full brightness while everything else dims, so you can see *where* on the two axes your matches sit. **Isolate** hides the rest entirely. Suggestion chips appear in a strip above the grid. ## Zooming and clusters When you zoom out far enough that dots would overlap, the Scatter thins each crowded area down to a few representatives. Those surviving dots grow to show they stand in for a cluster; as you zoom back in, they shrink and split into the individual samples behind them. Something not working? Check our [troubleshooting guide](/reference/troubleshooting) or reach out on [Discord](https://discord.gg/JcV3kEw3V7). # Link, Tempo and Key Lock Source: https://docs.samplevault.ai/core-features/link-tempo-key Synchronize playback with your DAW and lock samples to a global tempo and key ## Overview Sample Vault can time-stretch and transpose your samples in real-time so everything plays at the same tempo and key. Combined with Ableton Link, you can audition samples perfectly in sync with your DAW. ## Tempo Lock Tempo Lock time-stretches every sample to match a global BPM. When you browse loops at different tempos, they all play back at the tempo you set. ### How to Use Click the **tempo lock** button in the title bar. A BPM input field appears. Enter the BPM you want all samples to play at. You can type a value or use the arrow keys to adjust. Play any sample — it will be time-stretched to your target tempo in real-time. Tempo Lock only affects samples that have a detected tempo. One-shots and samples without tempo metadata play back at their original speed. ## Key Lock Key Lock transposes samples to a target musical key. When enabled, every sample with a detected key is transposed so it plays in the key you choose. ### How to Use Click the **key lock** button in the title bar. A key selector appears. Select the target key from the dropdown (e.g. C, D#, F). Samples are transposed in real-time to match your chosen key. Key Lock only affects samples that have a detected musical key. Drum loops, percussion, and samples without key metadata are not transposed. ## Ableton Link [Ableton Link](https://www.ableton.com/en/link/) is a protocol that synchronizes tempo and playback position across applications and devices on the same network. Sample Vault uses Link to keep sample playback in sync with your DAW. ### Supported DAWs Any DAW or application that supports Ableton Link will work, including: * [Ableton Live](https://www.ableton.com/en/manual/synchronizing-with-link-tempo-follower-and-midi/) * [Bitwig Studio](https://www.bitwig.com/userguide/latest/the_dashboard/) * [Logic Pro](https://support.apple.com/guide/logicpro/ableton-link-in-logic-pro-for-mac-lgcp528602e4/mac) * [Reason](https://docs.reasonstudios.com/reason13/synchronization-and-advanced-midi#synchronization-using-ableton-link) Refer to your DAW's documentation for how to enable Ableton Link. ### How to Use Turn on Ableton Link in your DAW's preferences or transport bar. The exact location varies by DAW. Click the **Link** button in the Sample Vault title bar. Sample Vault will automatically discover other Link-enabled apps on your network. Play any sample and it will loop in sync with your DAW's transport. Tempo Lock is automatically enabled when Link is active. ### What Happens When Link Is Enabled * **Tempo is synced**: Sample Vault adopts the shared Link session tempo. Any participant can change the tempo and all peers follow. * **Playback loops**: Samples loop continuously until you pause them, staying in sync with the shared timeline. * **Tempo Lock activates**: Tempo Lock is automatically enabled so all samples play at the shared tempo. Link works over your local network. Make sure Sample Vault and your DAW are on the same Wi-Fi or wired network. ## Combining Features Tempo Lock, Key Lock, and Link work together. A common workflow: 1. Enable **Link** to sync with your DAW session 2. Enable **Key Lock** and set the key of your track 3. Browse samples — everything plays in sync and in key This lets you audition samples from your browser as if they were already placed in your DAW project. ## Limitations * **Samples without metadata**: Tempo Lock and Key Lock only affect samples with detected tempo or key. Samples missing this metadata play back unmodified. ## Troubleshooting ### Link is not detecting my DAW If Sample Vault doesn't see your DAW (or vice versa) after enabling Link: 1. Make sure both applications are on the same local network (Wi-Fi or wired) 2. Toggle Link off and on again in Sample Vault 3. Toggle Link off and on in your DAW as well 4. Check that your firewall is not blocking Link's network discovery Link uses UDP multicast for peer discovery. Some corporate or public networks block multicast traffic. If you're in a studio with a managed network, ask your network administrator to allow multicast on the local subnet. ### Samples are out of sync with my DAW If playback drifts or doesn't align with your DAW's transport: * Make sure **Tempo Lock** is enabled (it activates automatically with Link) * Verify the sample has a detected tempo — samples without tempo metadata are not time-stretched * Try pausing and replaying the sample to resync ### Tempo Lock or Key Lock has no effect on a sample Tempo Lock and Key Lock only work on samples that have detected tempo or key metadata. If a sample plays back unmodified: * Check whether the sample has a BPM or key value in the Library * Samples without this metadata (short one-shots, sound effects) are not affected # Note Editor Source: https://docs.samplevault.ai/core-features/note-editor Write project notes, to-do lists, and ideas alongside your sample library ## Overview The Note Editor is a rich text editor built into Sample Vault. Use it to keep project notes, reference information, ideas, and to-do lists alongside your sample library — all in one place, without switching to another app. ## Writing Notes Notes are written in [Markdown](https://www.markdownguide.org/basic-syntax/). If you're not familiar with Markdown syntax, the guide covers everything you need. Common formatting: | Syntax | Result | | - | - | | `**bold**` | **bold** | | `*italic*` | *italic* | | `# Heading` | Large heading | | `## Heading` | Subheading | | `- item` | Bullet list | | `- [ ] task` | Unchecked to-do | | `- [x] task` | Checked to-do | | `` `code` `` | Inline code | ## To-Do Lists Create checkable to-do items by writing `- [ ]` at the start of a line: ``` - [ ] Find reference kicks for the intro - [ ] Layer the main pad with a C minor sample - [x] Export stems ``` Click any checkbox in the editor to toggle it between checked and unchecked. Previously, to-do lists were a separate feature. They are now consolidated into notes. ## Slash Commands Type `/` anywhere in a note to open the command menu. Slash commands let you insert formatted blocks without typing Markdown syntax manually. Common commands: | Command | Inserts | | - | - | | `/heading` | A heading block | | `/bullet` | A bullet list | | `/task list` | A to-do item | | `/divider` | A horizontal rule | | `/code` | A code block | ## Reordering Sections Drag the handle to the left of any block to move it up or down. This lets you reorganize your notes without cutting and pasting. ## Notes and Projects Notes are associated with your active project. Switching projects in the title bar shows the notes for that project. # Projects Source: https://docs.samplevault.ai/core-features/projects A board for the tracks you're working on, with the DAW sessions, notes, and bounces that belong to each one ## Overview A project is a container for a track you're working on. It carries the context that track needs — BPM, key, genre, mood, notes, and to-dos — and gathers the DAW sessions, ideas, and bounces that belong to it in one place. The assistant reads a project's context on every message, and [Recall](/core-features/recall) scopes your activity to the active project, so switching projects switches the frame of reference for everything. Projects also give you a board across everything in flight, so a track is never just an untitled session buried in a folder somewhere. ## The board Every project sits on a board, grouped by the stage it's in. Stages run in a fixed order from first idea through to a finished master, and you move a project along as the work progresses. The board is the answer to "what am I actually in the middle of" — the tracks you started, the ones stuck in the mix, the ones ready to bounce. ## Inside a project Open a project to its detail view, which brings together everything attached to that track: * **Sessions** — the DAW sessions that belong to the project (see [DAW projects](#daw-projects) below), including which one you most recently had open. * **Samples** — the sounds used across the project's sessions, gathered as playable rows so you can find and reuse them; anything a session references that isn't in your library is listed separately. * **Notes and to-dos** — the same [notes](/core-features/note-editor) and checklist that travel with the project's context. * **Bounces** — renders and reference mixes you keep against the project, so you can play back where a track was at any point without opening the DAW. * **Open in your DAW** — jump straight from a session back into the DAW that made it. ## DAW projects Connect the folders where your DAW keeps its project files and Sample Vault imports the sessions inside them, then keeps watching those folders for new or changed sessions. Ableton Live projects are supported first; connect as many folders as you like. Once your sessions are connected, two things become possible that a folder of files can't do on its own: * **Fit-aware recall** — Sample Vault knows the tempo and key of the session you're working in, and the character of the sounds already in it. Recall can lean on that, so "find me something that fits" means something concrete: sounds that sit in the same tempo, key, and sonic territory as the track in front of you. * **Reverse lookup** — for any sample, see the sessions it was used in. When you're wondering "where did I already use this," the answer is on the sample itself. Creating and organizing projects — the board, stages, notes, to-dos, and bounces — is free. Connecting your DAW project folders, and the session import, fit-aware recall, and reverse lookup that depend on it, is part of the paid plans. See [Plans & Usage](/reference/plans-usage) for what each plan includes. Everything here lives in your library's local database on your machine. # Search overlay Source: https://docs.samplevault.ai/core-features/quick-actions An always-on-top search bar that sits over your DAW, so you can find a sample and drop it into your session without switching windows ## Overview The search overlay is a slim, always-on-top bar you can summon over any window — your DAW included. Find a sample, refine it, and drag it straight into your session without leaving what you're doing. Even when the main Sample Vault window is out of the way, recall stays one keystroke away. ## When it appears By default the overlay surfaces whenever Sample Vault isn't the window you're working in, and steps aside when you bring the app back to the front — so it's there while you're in your DAW and gone when you're not. You can set it to stay on all the time, or turn it off entirely, and toggle it for the current session from the system tray. Choose the behavior in **Settings**; the exact path and options live in the [settings reference](/reference/settings). Summon it with the search overlay shortcut from anywhere, even when Sample Vault isn't focused. Escape steps back a level and then dismisses it. The shortcut is listed under [Keyboard Shortcuts](/reference/keyboard-shortcuts). ## Finding a sample Type and the overlay searches your library as you go, matching both by name and by how samples sound. Results audition automatically so you can step through them with the arrow keys — even while you keep typing — and drag the one you want directly onto a DAW track. When a result is close, use it as a reference to find nearby sounds without leaving the overlay; switch between overall, timbre, and frequency-band matching as the search narrows. It reads the same search tokens as the browser (`#120 @Cmin !loop`); see [Search Syntax](/core-features/search-syntax) for the full grammar. Narrow further with the same filters and saved presets as the library. ## Going deeper When an instant match isn't enough, send your query as a [deep search](/core-features/search-modes): describe a sound in your own words or ask about your library, keep refining in a short back-and-forth, and the picks land in the results below without opening the full app. Deep search and the assistant need a plan that includes them — see [Plans & Usage](/reference/plans-usage). Instant search and grabbing audio into Tape work even when you're signed out. ## Tape and sync within reach Grab the last few seconds of system audio into [Tape](/core-features/tape) without opening its window, so a sound you just heard is ready to drop into your session. The overlay also carries the same tempo and key sync — including Ableton Link — as the main app, so your BPM lock and transpose stay consistent everywhere you work. ## Recall the main window When you want the full browsing experience, one action brings the main Sample Vault window back to the front — handy for moving from a quick grab into deeper library work. **Audition without lifting your hands.** Arrow-key through results to play them one after another, then drag the keeper straight onto a track — ideal for finding the right kick or riser mid-session without switching windows. # Recall Source: https://docs.samplevault.ai/core-features/recall Track your activity and give the AI assistant context about how you work ## Overview Recall automatically tracks your interactions in Sample Vault — every search, drag, copy, Deep Search, and assistant message — and displays them in a timeline you can browse. It also feeds this history to the AI assistant so it can give you smarter, more relevant suggestions based on what you've actually been doing. ## Why Recall? Without Recall, the AI assistant treats every conversation as a blank slate. With Recall enabled, it knows that you've been browsing dark techno kicks, dragging hi-hats into your DAW, and searching for 140 BPM loops — so when you ask "find me something that fits", it already has context. **What it tracks:** * **Sample drags** — when you drag a sample into your DAW * **Sample copies** — when you copy a sample to your clipboard * **Sample browsing** — filter changes in the Library (search terms, tags, genres, tempo range, key, etc.) * **Searches** — searches by description, including Deep Search * **Assistant messages** — messages you send to the AI chat ## The Recall Panel Open the Recall panel from the **Layouts** menu or by adding it to any dock layout. It appears as a scrollable timeline of your recent interactions, newest first. ### Project Scoping Recall is project-aware. When you have a project active, the panel shows only interactions from that project. When no project is selected, it shows global (non-project) interactions. The current scope is displayed at the top of the panel. ### Filtering by Type Use the toggle buttons at the top of the panel to filter by interaction type. For example, show only "Dragged" and "Copied" events to review which samples you've recently used, or show only "Assistant" events to revisit your chat history. ### Sample Playback For drag and copy events, Recall doesn't just show the filename — it renders a full sample player inline. You can play back, drag, or copy the sample again directly from the Recall panel without navigating back to where you originally found it. ## AI Context Recall feeds a compact summary of your recent activity into the AI assistant's system prompt. This happens automatically whenever you send a message — you don't need to do anything. The AI receives up to 30 recent interactions, formatted as a brief activity log. This gives the assistant awareness of: * What samples you've been working with * What searches you've been running * What filters and genres you've been exploring * What you've asked the assistant previously This context helps the AI make better suggestions, avoid repeating itself, and understand the direction of your current session. Recall works best when you have a project active. The AI receives project-scoped history, so it understands the context of your current project rather than mixing in unrelated activity from other sessions. ## Privacy All interaction history is stored **locally on your machine** in Sample Vault's local database. It is never sent to any server or shared with anyone. The only place this data leaves your machine is as part of the AI assistant prompt when you actively send a message — and even then, it's a truncated summary, not the full history. # Library Source: https://docs.samplevault.ai/core-features/sample-browser Browse, play, and organize samples from your personal library ## Overview The Library is your main workspace for discovering and organizing the samples in your personal collection. ## Layout The Library has two main regions: * **Toolbar** with the [search bar](/core-features/search-syntax) (your master control for every filter) and the **Filters** button beside it * **Results** — the samples themselves, at full width, with any filters you've applied shown as chips above them Several related systems drive what you see in the results: * **[Deep Search](/core-features/search-modes)** — type a name or describe the sound; matches blend automatically, with an AI pass on demand * **[Search Syntax](/core-features/search-syntax)** — token grammar the search bar understands * **[Filters](/core-features/filters)** — the search bar (master control), the Filters dialog, and saved presets * **[Sound Profile](/core-features/sound-profile)** — sonic characteristics ## Views The results aren't limited to a list. Each Library panel can show its samples as a: * **List** — the default ranked rows, with full details and inline playback * **[Map](/core-features/library-map)** — a field where samples that sound alike sit close together * **[Circle of Fifths](/core-features/library-circle)** — every sample placed by musical key * **[Scatter Plot](/core-features/library-scatter)** — samples plotted on two sound characteristics you choose Every view honours the same search and filters, and each panel remembers its own view — so you can keep one panel on the list and another on the Map. ## Collections Collections let you group samples together for easy access — by project, mood, style, or anything you choose. They live in the **Collections** section of the [Filters](/core-features/filters) dialog. ### Creating Collections * Pick **New collection** in the Collections section to start one * Right-click any sample and choose **Add to collection** to drop it into an existing collection * Select a collection to filter results to only its samples ### Managing Collections * Rename a collection at any time — the samples inside it are unaffected * Delete a collection when you no longer need it; the samples stay in your library and are simply no longer grouped under it * Select several samples at once to add them all to a collection, or remove them from one they all belong to, in a single action ### Live Collections (paid plans) Describe the collection you want and Sample Vault builds it from your library, then keeps it current on its own. A live collection is organised into labelled groups — different facets of what you asked for — and each group fills automatically as matching samples arrive, with no manual upkeep. * Browse a live collection group by group, and open any group on its own to see everything it matches * Adjust a live collection by describing the change in plain language, rather than editing it by hand * Live collections update themselves and stay read-only; manual collections remain yours to curate by hand * Sample Vault also builds a starter set of live collections from your library on its own — informed by your producer profile and any standing instructions you've set — so there's something to browse before you describe your own Available on paid plans. ## Playback ### Player Controls * **Play/Pause**: Click the play button or press `2` * **Previous / Next**: Navigate through samples in the same folder using `1` / `3` * **Waveform**: A detailed picture of the sample, colored by frequency content so you can tell bass, mids, and highs apart at a glance. Click anywhere on it to seek to that position. * **Seek bar**: Where a row is too narrow to show a waveform, the playing sample grows a slim progress bar along its bottom edge. Click or drag it to move through the file. `Shift + ←` / `Shift + →` do the same from the keyboard, and repeat while held. ### Tempo & Key Lock * **Tempo Lock**: Time-stretches samples to a global BPM * **Key Lock**: Transposes samples to a target key * **Ableton Link**: Syncs with your DAW's transport See [Link, Tempo and Key Lock](/core-features/link-tempo-key) for details. Cloud samples download on first play and are then cached locally for instant playback. Local samples on your disk play immediately. ### File Integration * **Drag and Drop**: Drag samples directly into your DAW * **Copy to Clipboard**: One-click copy for pasting into audio software * **Show in Finder/Explorer**: Open the sample's folder on disk ## Working with Samples ### Sample Details Each sample row shows: * Sample name and folder path * BPM, key, and duration (when available) * Favorite button * Reveal in [File Browser](/core-features/file-browser) button ### Right-Click Menu Right-click any sample for quick actions: * Add to Collection * **Key** / **Scale** / **Type** — pick a new value from a submenu; applied instantly * **Tempo** — pick from your library's most common tempos or type a custom BPM * **Creator** / **Sample pack** — search your existing values or type a new one to add it * **Category** — reassign the sample to any category in the full tree * **Tags** / **Genres** — toggle chips to add or remove; changes save as you click * **Copy timestretched** — copy the sample with the current time-stretch applied (visible when tempo lock is active) * **Copy pitched** — copy the sample with key transposition applied (visible when key lock is active) * **Copy timestretched & pitched** — copy with both effects baked in (visible when both are active) ## Multi-select & batch actions The same shortcuts work in every list of samples — the Library, [Deep Search](/core-features/search-modes) results, [File Browser](/core-features/file-browser), [Similar Samples](/core-features/similar-samples), and sample chips in the [AI Assistant](/core-features/ai-assistant) chat. | Shortcut | Action | | - | - | | `Cmd/Ctrl + click` | Toggle a sample in the selection | | `Shift + click` | Range-select between two samples | | `Cmd/Ctrl + A` | Select every sample currently loaded | The Library pages results in batches — scroll to load more and press `Cmd/Ctrl + A` again to extend the selection. In the chat, `Shift + click` ranges within a single message, and with nothing focused `Cmd/Ctrl + A` falls through to text selection as you'd expect. ### The selection bar Any selection across any panel feeds the same bar at the bottom of the screen. It shows the current count, a **Clear** button, and a **More** menu that mirrors the directory right-click actions: Key, Scale, Tempo, Creator, Sample pack, Type, Category, Tags, Genres, Analyze, Classify, Enrich, Upload, Delete. Drag any selected entry into your DAW, Finder, or Explorer to drop the entire batch. Cloud-only samples cache in the background when you drag a multi-selection. Drop them again once caching finishes to include them in the dropped batch. ## File Browser To navigate your directory tree, add new sample directories, and control upload and analysis settings, see the [File Browser](/core-features/file-browser). ## Keyboard Shortcuts | Key | Action | | - | - | | `1` / `Down Arrow` | Play previous sample in folder | | `2` / `Spacebar` | Play / pause | | `3` / `Up Arrow` | Play next sample in folder | | `Cmd/Ctrl + click` | Toggle sample selection | | `Shift + click` | Range-select samples | | `Cmd/Ctrl + A` | Select all loaded samples | # Deep Tagging Source: https://docs.samplevault.ai/core-features/sample-enrichment How real genres, tags, creators, and sample packs get onto your samples ## Overview Every sample in your library gets labels for free: a category, some tags, a genre guess. Those come from the filename and the folder it sits in, which is fine when a pack is named well and useless when it isn't. **Deep Tagging** replaces those guesses with real ones. It reads your library in context — the folder a sample sits in, its siblings, the pack it came from — and works out the genre, the creator, and the sample pack for each file. That does two things: it makes the genre, creator, and pack filters trustworthy, and it sharpens [Deep Search](/core-features/search-modes), because the words being matched against are now describing the sound rather than repeating a filename. Deep Tagging is the only part of processing that needs an internet connection. Everything else, including search by description, works offline and is free on every plan. ## What it changes | | Without Deep Tagging | With Deep Tagging | | - | - | - | | Genre, creator, pack filters | Guessed from filenames, so patchy | Reliable across the folders you've tagged | | [Deep Search](/core-features/search-modes) | Works, but matches against filename-derived words | Matches against real descriptions of the sound | | [Similar Samples](/core-features/similar-samples) | Unaffected — matches on sound either way | Unaffected | | Tempo, key, [sound characteristics](/core-features/sound-profile) | Unaffected — measured from the audio | Unaffected | The point worth holding onto: Deep Tagging changes label **quality**, never whether a sample is searchable. A library that has never been deep-tagged is still fully searchable by name, by sound, and by description. ## Running it Deep-tag a folder from its right-click menu in the [File Browser](/core-features/file-browser) or in Sample Directories, from the **Deep Search** row in the Library status readout, or by ticking folders in the Add Directory dialog when you first point Sample Vault at them. A run works through your library in a few passes and can take a while on a large one. Watch it in the Library status readout, which names the pass it's on. It's available across your whole library on any paid plan, with no cap on library size. On Demo Room you can deep-tag one folder to see what it does to your results. See [Plans & Usage](/reference/plans-usage) for what each plan includes. ## Choosing which labels are allowed Control which labels can be assigned from the taxonomy controls in [Settings](/reference/settings): * **Genres** — detected genre classifications * **Tags** — descriptive labels like "dark", "punchy", "ambient" * **Creators** — detected sample pack creators or artists * **Sample Packs** — detected sample pack names Turn individual labels off and they won't appear on any sample, even where Deep Tagging detected them. If labels are landing that don't match your library, turn them off here rather than editing samples one by one. Your library reconciles on the next pass. ## Organizing Your Library As your library grows, Sample Vault keeps it organized for you. A single pass tends three things together: it corrects and discovers categories by how samples sound, tidies your genre and tag lists, and assembles [live collections](/core-features/sample-browser). It runs on its own in the background, and you can start it on demand from **Settings** (see [Organization settings](/reference/settings)). Anything you've made or named by hand is left untouched, and the whole pass is available on paid plans — see [Plans & Usage](/reference/plans-usage) for what each plan includes. ### Categories by sound Sample Vault groups samples that sound alike and uses those groups to put each one in the right category: a sample whose category doesn't match how it sounds moves to the one its neighbours share, and a distinct kind of sound with no existing home — like hardstyle kicks or reverse cymbals — becomes its own category. Categories you've created yourself are never changed. ### Genres and tags As you deep-tag more folders the genre and tag lists grow, and left alone they sprawl into near-duplicates, one-off labels, and tags that only repeat something you can already filter by. Sample Vault merges duplicates, groups specific values under broader families, and drops tags that only echo another filter like a category, key, or tempo. The result is the family grouping you browse in the [filters](/core-features/filters). Tags you've created or renamed yourself are never deleted — organizing can only nest or relabel them, so your own vocabulary stays intact. ### Reviewing what changed Because these passes run on their own, Sample Vault keeps a record of the recent ones so you're never left guessing what moved. Each entry shows when the pass ran, whether it was automatic or one you started by hand, and exactly what it changed — the categories it named, the genres and tags it merged or grouped, and the collections it built. You'll find this history alongside the organization controls in **Settings**. # Deep Search Source: https://docs.samplevault.ai/core-features/search-modes Search by name and by how samples sound — then escalate to an AI pass when you want more precise matches ## Overview The Library has a single search bar with no modes to choose between. As you type, it searches two ways at once and blends the results: it matches your text against filenames and [tokens](/core-features/search-syntax), and it matches the *meaning* of your words against how each sample actually sounds. Exact hits and close-in-character matches come back together in one ranked list, updated as you type. This everyday search is free, works offline, and never leaves your machine. When you want sharper results for a tricky description, run **Deep Search**. It reads your request the way a person would — including the parts plain search struggles with, like exclusions ("no vocals") or nuance ("warm, nothing cheesy") — and turns it into a search plus a set of filters, checking the results and adjusting before handing them back. So you can: * **Look something up by name** — type part of a filename and the exact matches surface first. * **Describe a sound** — type `dark atmospheric pad` or `punchy vintage kick` and samples with that character come back, even when their filenames say nothing useful. * **Mix both** — names, tokens, and plain description all live in the same query. Describing a sound is what makes search work across a library full of cryptically-named files — the filename isn't the sound. A sample becomes reachable this way once it's been processed far enough to know what it sounds like; see [Adding & Managing Samples](/get-started/adding-samples) for the stages. ## Deep Search The everyday search runs the moment you type and costs nothing. **Deep Search** is the optional next step: it interprets what you asked for and sets up the search and filters to match, instead of leaving you to translate a vibe into filters by hand. Whatever it decides shows up as the normal, editable filters — so you can see why a result came back and change anything it chose. Use it when: * The quick results are in the right area but the best matches aren't at the top. * Your request has an exclusion or a subtlety that a straight description misses. * You're searching by feel across poorly-named samples and want the heavy lifting done for you. While it runs, Deep Search shows what it's trying, and when it finishes it leaves a short note about what it applied — including when your library doesn't contain what you asked for and it picked the nearest thing instead. Exclusions land as removable `not:` chips, and preferences ("ideally darker") as removable `~` chips that lean the ranking without hiding anything. ### Start from a sample When you've found something *close*, you can search from it instead of describing it from scratch: right-click a sample, choose to deep-search from it, and describe only the shift — "darker, more sub, less clicky". The search anchors to that sample's measured character and moves it the way you asked, so the results are variations of a sound you already like rather than fresh guesses. Deep Search needs an internet connection and draws on your plan's shared **AI actions** allowance — the same pool the assistant chat, Sketch Designer, and library suggestions use. Free accounts get a weekly allowance that refills each week; paid plans run Deep Search under fair use. When the allowance runs out, the free blended search keeps working as normal — see [Plans & Usage](/reference/plans-usage#starter-pools) for how the allowance works. ## Writing effective descriptions When you search by character rather than by name, describe the **character, role, and vibe** you're after and let tokens and filters handle the exact numbers. * **Mood and character**: `dark and atmospheric`, `warm and nostalgic`, `aggressive and gritty` * **Instrument or role**: `trap hi-hats`, `analog bass line`, `ambient pad for a breakdown` * **Production style**: `lo-fi dusty drums`, `vintage tape saturation`, `clean modern synth` * **Adjectives** that describe timbre: warm, punchy, smooth, gritty, airy, metallic Keep a description **under 100 characters** — the search bar enforces this limit. Don't pack BPM, key, or sample type into the words — pin those down with the [Filters](/core-features/filters) dialog or [search tokens](/core-features/search-syntax) like `#120 @Cmin !loop` instead, and reserve the description for purely descriptive language. Tokens act as hard filters on top of the match, so `dark atmospheric pad #120 @Cmin` locks you to 120 BPM in C minor while the words find the vibe. ## Search suggestions Before you type, you'll see **suggestion chips** drawn from your search history — pick one to run that search immediately. For a fresh direction, open prompt ideas to see suggested searches based on your library; choose one to run it, or dismiss the ideas without changing your search. You can ask for another set whenever you like. As soon as you start typing, a **suggestion dropdown** opens with matching tags, genres, creators, sample packs, and collections, so you can insert the right token without remembering its prefix. See [Search Syntax](/core-features/search-syntax) for the full grammar and dropdown navigation. ## Combining tokens with search The same token grammar works in every query — `#120`, `@Cmin`, `!loop`, `$techno`, `%dark`, `&adam`, `+cymatics`, `*` (favorites), `category:`, `collection:`, `sort:`, plus sound-profile ranges like `atk:30-80`. Tokens are always **hard filters** that narrow the result set exactly; the remaining plain text drives the blended name-and-meaning match, whether you run the quick search or Deep Search. See [Search Syntax](/core-features/search-syntax) for the full reference. ## Multi-select in results `Cmd/Ctrl + click`, `Shift + click`, and `Cmd/Ctrl + A` all work on search results. Selection feeds the same bar used everywhere else, so you can drag a batch of results into your DAW or bulk-edit through the **More** menu. See [Multi-select & batch actions](/core-features/sample-browser#multi-select-batch-actions). ## Troubleshooting ### No results If a search returns nothing, try: * **Simplify** — fewer, more common terms * **Reword** — try "synth" instead of "synthesizer", "drums" instead of "percussion" * **Loosen specifics** — drop very narrow BPM or key requirements, or move them into [tokens](/core-features/search-syntax) where they act as filters * **Broaden the genre** — "electronic" instead of "future bass" * **Try Deep Search** — when the quick search is close but the ranking's off, the AI pass often surfaces matches the instant search misses The "No samples found" screen shows your exact query so you can refine it. ### Why don't I see all my samples in the results? Search ranks by how closely each sample matches what you typed rather than listing everything. To browse your whole library, clear the search bar or narrow down with the [filters](/core-features/filters) instead. ### A sample won't come up by description Describing a sound only works once a sample has been processed far enough to know what it sounds like. Until then it's still reachable by name and tokens. Check the **Library status** panel — see [Adding & Managing Samples](/get-started/adding-samples) for what each stage unlocks. Detail matters too: a sample you haven't [deep-tagged](/core-features/sample-enrichment) still matches by sound, but less sharply, so on a large library that's mostly not deep-tagged the right sound can rank low. When a search comes up thin for that reason, the results say how much of your library is tagged in detail and point you to deep-tag more. ### Character limit Descriptions are capped at 100 characters. The search bar warns you when you go over. Trim to the most important descriptive words and pin specs (BPM, key, type) down with [tokens](/core-features/search-syntax) instead. # Search Syntax Source: https://docs.samplevault.ai/core-features/search-syntax Token grammar for filtering by tag, genre, creator, pack, category, key, BPM, type, sound profile, and more from the search bar ## Overview The search bar is your master control. Every filter — tags, genres, creators, sample packs, key, BPM, sample type, favorites, categories, collections, sort order, sound profile characteristics — can be expressed as a typed token. Tokens render as colored chips inside the input. Combine them freely with each other and with plain text. You don't need to memorize the syntax. Start typing and the [smart suggestions dropdown](#smart-suggestions) shows the right token automatically — picking a suggestion inserts it for you. ## Tokens Most dimensions have a short single-character prefix (the high-frequency ones, optimised for typing speed) and a long-form `field:value` alternative (self-documenting, autocomplete-driven). | Dimension | Short prefix | Long form | Example | | - | - | - | - | | Tag | `%bass` | `tag:bass` | `%dark` `%"sci fi"` | | Genre | `$dnb` | `genre:dnb` | `$techno` `$"drum and bass"` | | Tempo | `#120` | `bpm:120` | `#90-140` `#100-` `#-130` | | Key + scale | `@Cmin` | `key:Cmin` | `@C` `@Cmaj` `@F#min` | | Sample type | `!loop` | `type:loop`, `is:loop` | `!loop` `!oneshot` `is:oneshot` | | Favorite | `*` | `is:favorite` | `*` | | Creator | `&adam` | `creator:adam` | `&"Adam Smith"` | | Sample pack | `+cymatics` | `pack:cymatics` | `+"Cymatics Hyperbits"` | | Category (virtual directory) | — | `category:drums/kicks` | `category:bass` | | Collection | — | `collection:wip` | `collection:"Work in Progress"` | | Sort order | — | `sort:newest` | `sort:tempo-asc` `sort:relevance` | | Sound profile (16 characteristics) | — | `attack:30-80` etc. | `bright:50-100` `stereo:80-100` | | Plain text | (anything else) | — | `punchy kick` | Multi-word values use double quotes: `%"drum and bass"`, `&"Adam Smith"`, `collection:"Work in Progress"`. ### Sound profile aliases The 16 sound profile characteristics each accept a short alias and the canonical name. Values are percent ranges from 0 to 100. | Alias | Full name | Example | | - | - | - | | `attack` / `atk` | Attack | `atk:30-80` | | `bright` / `brightness` | Brightness | `bright:0-50` | | `bass` | Bass content | `bass:60-100` | | `sub` / `subbass` | Sub bass | `sub:50-` | | `mids` | Mids | `mids:40-70` | | `highs` | Highs | `highs:60-100` | | `lomids` / `lowmids` | Low mids | `lomids:30-60` | | `stereo` | Stereo width | `stereo:80-100` | | `perc` / `percussiveness` | Percussiveness | `perc:50-100` | | `dyn` / `dynamic` | Dynamic range | `dyn:30-80` | | `pitch` | Pitch stability | `pitch:60-100` | | `rhythm` | Rhythmic density | `rhythm:50-` | | `tonality` | Tonality | `tonality:70-100` | | `noise` | Noise | `noise:0-30` | | `complexity` | Complexity | `complexity:60-` | | `temporal` | Temporal centroid | `temporal:30-70` | Range syntax matches BPM: `atk:30-80` (between), `atk:30-` (at least), `atk:-80` (at most), `atk:50` (exact). ## Stacking tokens Tokens stack freely with each other and with plain text: ``` #120-130 @Cmin !loop %dark $techno &adam atk:50-100 punchy kick ``` This finds 120–130 BPM loops in C minor, tagged "dark", genre techno, by creator "adam", with attack at 50–100%, with the keyword "punchy kick" in the filename. ## Smart suggestions Type two characters and a suggestion dropdown opens with matching values grouped by dimension — sample type (loop / one-shot), categories from your library folders, tags, genres, creators, sample packs, collections, plus a **Filters** section that suggests field shortcuts (`creator:`, `pack:`, `collection:`, etc.) when you start typing one. So `loop` or `one-shot` will offer the matching type filter, and a folder name like `kick` or `drums` will offer the matching category — no need to know the prefix syntax. Once you've committed to a dimension by typing its prefix or field, the dropdown narrows to that dimension and shows results immediately (no two-character gate inside an explicit token): * `%dr` — only tags matching "dr" * `&` — top creators by sample count, no prefix needed * `category:` — full category list **Keyboard:** | Key | Action | | - | - | | `↓` / `↑` | Move highlight | | `Tab` | Autocomplete the highlighted suggestion | | `Enter` | Apply highlighted suggestion, or run the search if none is highlighted | | `Esc` | Close the dropdown without applying | ## Removing chips Selected filters live as colored chips inside the search bar. **Click any chip to remove it** — hovering previews the removal with a red strike-through and × icon. To clear everything at once, click the **×** button on the right side of the input. ## How tokens and text work together Tokens are always **hard filters** — they narrow the result set exactly, whatever else you type. The remaining **plain text** drives the search: it's matched against filenames *and* against how each sample sounds at the same time, with the closest matches ranked first. Filters apply live as you type. This is the way to combine a description with technical constraints: `dark atmospheric pad #120 @Cmin` locks results to 120 BPM in C minor while the words find the vibe. The same token grammar powers the [search overlay](/core-features/quick-actions) — anything you learn here works there too. # Similar Samples Source: https://docs.samplevault.ai/core-features/similar-samples Automatically discover samples that match what you're playing ## Overview The Similar Samples tab surfaces the most sonically similar samples in your library whenever you play a sample — no searching required. ## How It Works Play any sample from anywhere in the app and the Similar Samples tab updates automatically with the closest matches. Similarity is based on a **semantic understanding of the sound itself** — an AI classification embedding produced during the [Classified](/get-started/adding-samples#the-processing-pipeline) stage that captures what the sample actually sounds like, not just its tempo, key, or filename. Two samples with very different metadata can rank as highly similar if they share the same sonic character. Keep the Similar Samples tab open alongside the Library. Every sample you audition becomes a starting point for finding related sounds. ## Using Similar Samples 1. Open the **Similar Samples** tab in your layout 2. Play any sample from anywhere in the app 3. The tab immediately populates with the closest matches from your library Each result includes the same playback controls and context menu as the main Library — audition, favorite, drag to your DAW, or open in the AI chat without switching tabs. The standard [multi-select shortcuts](/core-features/sample-browser#multi-select-batch-actions) work here too: pick a batch of similar samples and drag the whole set into your DAW or bulk-edit through the selection bar. ## Similarity Scoring Results are ranked by similarity score. A higher score means the sample sits closer to what you played in the AI's understanding of sound — sharing the same sonic character, texture, and instrumentation rather than matching on filename or tags. Samples have to reach the **Classified** stage before they appear in Similar Samples. If a sample is missing from results, check the Library status panel — it may still be queued for classification. Right-click the directory and choose **Classify samples** to prioritize it. # Sketch Source: https://docs.samplevault.ai/core-features/sketch Build beats from your own samples with a polymetric pad grid, then export to your DAW ## Overview Sketch turns your library into a drum machine. You assign your own samples to pads, program each pad on its own step lane, and play the whole thing back in sync. It's made for sketching a groove and handing it to your DAW — not for finishing a track inside Sample Vault. Playback and sample suggestions run on your machine, so Sketch works offline and is free to use. ## The pad grid Each pad holds one sample and plays one voice. You choose how many pads you want, from a single row up to an 8×8 grid, and every pad is mapped to a musical note so it can be played from a MIDI controller. To fill a pad you can pick a sample from your library, drop one onto it, or let Sketch choose. Clicking a pad auditions whatever is loaded. Each pad can be muted, soloed, or locked. **Locking** a pad protects its sample from the random, reroll, and kit-building actions, so you can keep the sounds you like and reroll the rest. ## Polymetric step lanes Every pad gets its own step lane, and lanes are independent: each has its own number of steps and its own rate. Give one pad seven steps and another sixteen and they drift against each other, only realigning after a long cycle. That polymetry is the point — it keeps a loop evolving instead of repeating every bar. Within a lane you can: * Set how many steps it has and how fast they run, including triplet and dotted timings. * Set how hard each step hits by dragging it up or down. * Tune individual steps up or down, with the resulting note shown against the pad's key. * **Snap to key** so step pitches stay inside the detected key and scale and never clash. * Put pads in a **choke group** so they cut each other off — the way a closed hat silences an open one, or each note of an 808 glide stops the last. ## Sourcing samples Because Sample Vault already knows what your samples sound like, Sketch can help you fill pads: * **Swap for similar** steps a pad through samples that sound like the one it holds, so you can audition variations without leaving the grid. * **Build kit** fills the unlocked pads with a matching set of sounds pulled from your library. * **Reroll** swaps every unlocked pad at once — either for fresh picks from each pad's own source or for new similar-sounding samples. * **Random** drops a random sample from your library onto a pad. * **Sketch a beat** builds a whole groove from a short description — choosing the kit, programming the pattern, setting the tempo and swing to fit, and layering in loops under the grid where they suit the request — or reworks what's already loaded when you describe a tweak. Locked pads and loops stay untouched. * **Design this lane** points the same designer at a single pad to swap its sound, rework its rhythm, or both, leaving the rest of the grid alone. ## Playing along The whole pattern shares one tempo and one swing amount. When an Ableton Link session is running, Sketch follows the shared tempo so it stays locked to your DAW. See [Link, Tempo and Key Lock](/core-features/link-tempo-key) for how Link works. Connect a MIDI controller to play the pads by hand: each pad responds to its mapped note, and how hard you hit translates into how hard the pad plays. Select your controller from Sketch's controls once it's plugged in. Press record to start rolling and capture a performance straight into the pattern. Each hit lands on the nearest step of its pad's lane, at that lane's rate, and your playing dynamics are kept. Recording adds to what's already there, so you can build a groove pad by pad over several passes. Patterns save automatically as you work, so you can close the panel and pick up where you left off. ## Exporting to your DAW When the groove is right, take it with you — or keep it in your library: * **MIDI clip** exports the pattern as a single MIDI clip with each pad on its own note, alongside a manifest that maps each note to the sample it should play, so your DAW can rebuild the kit. * **Audio render** bounces the pattern to a WAV file. Because polymetric patterns can run for a very long time before they repeat, the render is clamped to a sensible length. * **Save as a sample** renders the pattern back into your library as an audio file — analyzed on the spot and tagged as a loop at the pattern's tempo — so you can find it, audition it, and drop it onto a pad later. * **Copy to clipboard** puts the MIDI clip or the WAV render on your clipboard, ready to paste straight into your DAW or a folder. * **Collect kit sounds to folder** copies every sound the kit uses into one folder you pick, so the whole kit lands in a single place on disk. The originals stay where they are — this gathers the files you already own, it doesn't move them. * **Copy kit sounds to clipboard** puts every sound in the kit on your clipboard at once, ready to paste into your DAW or a folder. ## Limitations * **Swap for similar needs analyzed samples.** Pads filled from raw files outside your scanned library can still play, but won't have similar-sound matches until the sample is in your library. * **Built for sketching, not arranging.** There's no song mode or arrangement timeline — build a pattern, then finish the arrangement in your DAW. # Sound Profile Source: https://docs.samplevault.ai/core-features/sound-profile Filter samples by sonic characteristics using range sliders ## Overview The **Sound Profile** section of the [Library](/core-features/sample-browser) [Filters](/core-features/filters) dialog lets you filter samples by their sonic characteristics. Instead of browsing by tags or genres, you can narrow results based on measurable audio properties extracted during [analysis](/get-started/adding-samples). ## How It Works Each characteristic is represented as a **range slider** with a minimum and maximum value. Set a range on any characteristic to filter your results to only samples that fall within that range. * Drag the left handle to set a minimum threshold * Drag the right handle to set a maximum threshold * Active filters appear as **removable chips** above the results * Click the **X** on any chip to remove that filter You can combine multiple Sound Profile filters with each other and with other Library filters (genre, key, BPM, tags) for precise results. ## Available Characteristics ### Core | Characteristic | Description | | - | - | | **Tonality** | How tonal vs. noise-based the sample is | | **Percussiveness** | How percussive vs. sustained the sample sounds | | **Brightness** | Presence of high-frequency energy relative to overall spectrum | | **Noisiness** | Amount of noise content in the signal | | **Complexity** | Spectral and rhythmic complexity of the audio | | **Attack** | How sharp or gradual the onset is | | **Rhythmic Density** | How many rhythmic events occur over time | | **Pitch Stability** | How consistent the pitch is throughout the sample | ### Frequency Content | Characteristic | Description | | - | - | | **Sub Bass** | Energy in the sub-bass range (roughly 20-60 Hz) | | **Bass** | Energy in the bass range (roughly 60-250 Hz) | | **Low Mids** | Energy in the lower midrange (roughly 250-500 Hz) | | **Mids** | Energy in the midrange (roughly 500 Hz-2 kHz) | | **Highs** | Energy in the high-frequency range (roughly 2 kHz+) | ### Stereo & Temporal | Characteristic | Description | | - | - | | **Stereo Width** | How wide the stereo image is (mono to fully spread) | | **Temporal Centroid** | Where the energy is concentrated in time (early vs. late) | | **Dynamic Range** | Difference between the quietest and loudest parts | ## Tips Sound Profile filters work best on samples that have completed audio analysis. If you've recently added samples, wait for analysis to finish before filtering by characteristics. * **Finding kicks**: Set high Percussiveness, high Attack, high Sub Bass * **Finding pads**: Set high Tonality, low Percussiveness, high Stereo Width * **Finding loops vs. textures**: Use Rhythmic Density to separate rhythmic loops from ambient textures * **Finding clean samples**: Set low Noisiness and high Pitch Stability # Suggestions Source: https://docs.samplevault.ai/core-features/suggestions Samples worth a listen based on what you've been working on ## Overview The Suggestions panel surfaces samples from your library based on what you've been doing in the session — recent searches, drags, the sample you're playing, and the project you're working on. It's a faster way to keep ideas moving when you don't know what to search for next. ## How It Works Suggestions arrive in two layers: * **Recent moves** (always-on): a row of samples acoustically similar to the last one you dragged or copied. Runs locally with no network call, so it's there immediately when you open the panel. * **AI-picked sections** (2–4 cuts per refresh): the model looks at your recent activity, your project's tempo and key, and what's in your library, then invents section titles and queries on the fly — *"A bass for the drop"*, *"Breakdown drums at 138"*, *"Hats to go with kick\_punchy.wav"*. Each cut picks the right search method for the job (vibe-based semantic search, anchor-on-a-sample similarity, or hard filters like BPM range and key). Sections that don't have decent matches in your library are quietly dropped before they reach the panel. The model is grounded on a snapshot of your library — top tags, top genres, BPM and key distribution — so suggestions stay in the territory your samples actually cover. ## When It Refreshes Suggestions refresh on real signals, not on a timer: * You open the panel * You commit a new search * You drag a new sample into your DAW * You switch projects * You click the refresh button Refreshes are debounced so rapid actions don't burn calls. The manual refresh button always fires immediately. ## Adding the Panel The Suggestions panel auto-mounts in the **Discovery** layout. To add it to another layout: 1. Open a New Tab in the dock 2. Pick **Suggestions** from the panel list 3. Drag it where you want it You can only have one suggestions panel open at a time. ## Plans and Limits The deterministic **Recent moves** section works on every plan without using any AI quota. The AI-picked sections call the LLM: * **Home Studio / Professional Suite** — counted against the same fair-use limits as the other AI features. * **Demo Room / Lifetime (no subscription)** — drawn from the shared starter **AI actions** pool that also covers Deep Search, the assistant chat, and the Sketch Designer. Suggestions are a passive assist: when the pool is spent they quietly stop rather than charging you, the panel keeps showing Recent moves, and it prompts you to upgrade or — on Lifetime — [add an OpenRouter key](/reference/settings#api-keys). Pair the Suggestions panel with the Library and Similar Samples in the same layout. Audition something interesting from Suggestions, then dig further with Similar Samples — and your next set of suggestions will shift based on what you just dragged. # Tape Source: https://docs.samplevault.ai/core-features/tape A rolling recorder that keeps the last few minutes of your system audio so you can grab a sound after it played ## Overview Tape continuously buffers what your computer is playing, so a sound you liked is still there after it stops — a synth patch you were tweaking, a sample preview that went by, or a happy accident you didn't catch in the moment. Instead of remembering to hit record before the good part, you reach back and keep it afterward. Capture runs on your machine. Audio stays local and nothing is uploaded. ## How it works While background capture is on, Tape keeps a rolling window of recent system audio — the most recent stretch is always available, and older audio falls out of the window as new audio comes in. Nothing is saved to your library until you choose to keep a piece of it. You interact with the buffer in two ways: * **Quick grab** — a global shortcut copies the most recent few seconds to your clipboard from anywhere, even while you're inside your DAW, so you can paste it straight in. This is the fast path when you know the sound just played and want it immediately. * **Scrub and select** — open Tape to see the buffer as a waveform, scroll back through it, and drag out an exact selection to play and keep. ## Marking a moment When something good plays and you want to find it later, drop a mark. Marks pin that point on the timeline so you can jump straight back to it instead of scrubbing to hunt for it. A mark stays available for as long as its audio is still in the buffer; once the window moves past it, the audio is gone and the mark retires with it. ## Keeping a grab Once you've picked a selection (or used quick grab), you can: * **Save it to your library** as a normal sample. Saved grabs go into their own Tape collection and are analyzed and tagged like any other sample, so they're immediately searchable. * **Copy it** to paste into your DAW. * **Drag it** straight onto a track. ## Settings Background capture is off until you turn it on. In the app's preferences you can: * Turn capture on or off. * Choose which output Tape listens to. * Set the buffer length — longer means you can reach further back, at the cost of more disk while capturing. Changing the length keeps what you've already captured, and the disk each length uses is shown so you can choose deliberately. * Set the quick-grab length and the global shortcut that copies it. An indicator appears while capture is running so it's always clear when Tape is listening. ## Plans Tape is part of the paid plans. Without one it still opens on its timeline, so you can scrub, select a region and see how a grab is made. Capturing your own audio and keeping a grab are what a paid plan adds. ## Permissions macOS asks for permission to record this computer's audio the first time you switch capture on, and Tape needs a recent version of macOS to record system output at all. If capture can't start, Tape says so rather than sitting idle. Quick grab can also paste for you once the sound is on your clipboard, which macOS treats as a separate permission. Without it the grab still lands on your clipboard and you paste it yourself. Tape is a desktop feature. It captures the audio your computer is playing, so anything routed through your system output — your DAW, a browser, a synth — can be grabbed. # Waveform Visualizer Source: https://docs.samplevault.ai/core-features/waveform-visualizer Interactive waveform display with zoom, seek, and tempo sync ## Overview The Waveform Visualizer panel provides a detailed, interactive view of the currently playing sample. Open it as a tab in your [dock layout](/core-features/dock-layouts). ## Controls * **Seek**: Click anywhere on the waveform to jump playback to that position * **Zoom**: Scroll to zoom in and out — zoom is anchored to your cursor position * **Pan**: Shift+scroll (or horizontal trackpad scroll) to pan left and right across the waveform ## Regions Drag across the waveform to select a region. The region plays on its own — playback starts at its beginning and stays inside it, so you can audition a hit or a phrase in isolation. Drag either edge to adjust it. Clicking inside the region seeks within it; clicking outside clears it and returns to full-track playback. Drag a region out of the panel to drop it straight onto a DAW track as an audio file, bounced at the sample's own rate and bit depth. The region stays selected, so you can keep auditioning or drop it again. Right-click a region to act on it: * **Copy to clipboard**: Copies the region as an audio file, ready to paste into your DAW or file manager * **Save in Sample Vault**: Bounces the region into your library as a new sample under a **Captures** source, analyzed and searchable like any other sample Dragging, copying, and saving regions are available in the desktop app. ## Tempo Sync Enable **[Tempo Sync](/core-features/link-tempo-key)** to replace the default time ruler with beat and bar markers aligned to the sample's detected BPM. This makes it easy to see musical structure — bars, beats, and subdivisions — directly on the waveform. Tempo Sync requires the sample to have a detected BPM. Samples without tempo metadata will display the standard time ruler. ## Rendering The visualizer uses a multi-LOD (level of detail) WebGL2 rendering pipeline. At high zoom levels, sinc interpolation preserves accurate waveform shape down to individual samples. The display stays smooth whether you're viewing the full waveform or zoomed in to inspect transients. # X-Ray Source: https://docs.samplevault.ai/core-features/x-ray Drop in a reference track and find the sounds in your own library that match it — as a whole, or split into stems ## Overview X-Ray turns a track you like into a way to search your own library. Drop in any track, loop, or sample, select the part that caught your ear, and X-Ray finds the closest sounds you already own. It's search by reference: instead of describing a sound in words, you point at one. It runs on your machine. The reference never leaves it, and nothing is uploaded. ## Opening X-Ray Open X-Ray as a panel, or right-click any sample and choose **Open in X-Ray**. It accepts common audio files — WAV, MP3, AIFF, and FLAC — so a bounce, a loop, or a favourite record all work as a reference. ## Matching against your library Select any region of the reference and X-Ray ranks the sounds in your library by how closely they match it. From there you can: * Play back and compare matches against the region you selected. * Save a matched region into your library as a new **X-Ray Captures** source. * Copy a match to the clipboard, or drag it straight into your DAW. * Drag the selected region itself out of the lane and onto a DAW track, bounced from the source you selected it on. ## Stem splitting Baseline X-Ray matches the reference as a whole. Stem splitting takes it further: it pulls the reference apart into **vocals, drums, bass, and everything else**, so you can match each part on its own. Drag a region on any stem lane and X-Ray searches just that part of the mix — a drum lane finds your drums instead of the whole track at once. Separation runs on your machine like the rest of X-Ray. The first time you split a track, a one-time model file downloads; after that it works offline. Baseline X-Ray — dropping in a reference and matching the whole mix — is free on every plan. Stem splitting is part of the paid plans. See [Plans & Usage](/reference/plans-usage) for what each plan includes. # AI Assistant FAQ Source: https://docs.samplevault.ai/faq/ai-assistant Frequently asked questions about the AI Assistant ## Overview ## How does the AI know what samples I have? The AI has access to every sample in your processed library. Add directories under **Settings → Sample Directories**; once they finish [analysis and deep-tagging](/get-started/adding-samples), they're searchable from chat. ## What's the difference between projects and regular chats? Projects carry context — BPM, key, genre, mood, notes — that the AI reads on every message. Regular chats are blank-slate. Use a project per track or session; use chats for general questions. ## What if the AI can't find what I'm looking for? The AI can only search samples that have completed processing. Check **Settings → Sample Directories** that your folders are scanned and deep-tagged. If they are, try alternate descriptions — the AI matches on how a sound *feels*, not on filename. ## Does the AI remember previous conversations? Within a conversation, always. Across sessions, the assistant also keeps [Memory](/core-features/ai-assistant#memory) — durable facts like your preferences and conventions, plus lighter context about your current work that fades if it stops coming up. You can see, edit, or switch all of it off in Settings. Cross-session memory comes with Professional Suite or Lifetime with your own key; every plan still gets your project context and recent activity. Each project keeps its own context, so switching projects switches the AI's frame of reference. ## What should I do if I hit my usage limit? Demo Room and Lifetime users without a subscription get a free weekly allowance of AI actions — shared across the AI features like chat, Deep Search, and the Sketch Designer — plus one free folder of Deep Tagging. The weekly allowance refills each week; to keep going without waiting, upgrade to a paid plan or, on Lifetime, configure your own AI key (BYOK). On paid plans Deep Tagging covers your whole library with no cap. The Home Studio assistant has a monthly allowance; on Professional Suite the assistant, Deep Search, AI Collections, and library organization all run under fair use. Hover the badge in the navigation bar or open **Settings → Billing & Usage** to see what's left. See [Plans & Usage](/reference/plans-usage) for full details. ## What documentation does the AI have access to? The complete Ableton Live manual — setup, instruments, effects, routing, recording, performance, and troubleshooting. The AI can search it semantically, so technical-term-imprecise questions still find the right section. More DAW manuals may be added later. ## What's the difference between 'Add to Collection' and 'Add to Favorites'? Favorites is a personal quick-access list. Collections are reusable groupings tied to a project or theme. A sample can be both favorited and in multiple collections. # General Questions Source: https://docs.samplevault.ai/faq/general Frequently asked questions about Sample Vault ## Overview ## What is Sample Vault? Sample Vault is a sample browser for music producers. It scans your existing folders, analyzes every file locally to extract tempo, key, and sound characteristics, then layers AI search and tagging on top so you can find sounds by describing them. ## What platforms does Sample Vault support? * Windows * Mac (Apple Silicon) * Mac (Intel) * Any modern browser (Chromium, Firefox, Safari) for paid plans — see [Browser Access](/reference/browser-access) ## Can I use Sample Vault from a browser? Yes, on Professional Suite. Sign in at [samplevault.ai](https://samplevault.ai) and click **Open app**. The Library, Deep Search, Sound Profile, Similar Samples, and AI Assistant all work against your cloud library. Library scanning, sample analysis, Deep Tagging, and Ableton Link still need the desktop app — they surface "desktop app required" prompts in the browser tab. See [Browser Access](/reference/browser-access) for the full breakdown. ## Does Sample Vault work offline? Yes. Browsing, filtering, the everyday search, audio playback, Sound Profile, and Similar Samples all run locally and work fully offline. The AI Assistant and Deep Search require internet. Sign in once while online before going offline. Changes made offline sync to the cloud when you reconnect (paid plans only). ## Does Sample Vault generate sounds using AI? No. Sample Vault doesn't generate audio. It analyzes the samples you already own and helps you find them. ## Does the AI assistant understand the contents of my samples? The AI works from filename, folder structure, and the audio features extracted during analysis (tempo, key, spectral profile). It doesn't listen to samples directly. For most production sample libraries this is enough, but unusual or poorly-named files can be misclassified. ## How does the AI analyze my samples? Local analysis extracts tempo, key, and a set of sound characteristics (brightness, attack, stereo width, etc.) from every file. Deep Tagging then assigns genre, mood, instrument, creator, and sample-pack labels based on filenames, folder structure, and audio content. See [Adding & Managing Samples](/get-started/adding-samples) for the full pipeline. ## What are collections in Sample Vault? Collections are custom groupings of samples — by project, mood, or anything you choose. On paid plans, you can describe a collection and have Sample Vault build it from your library and keep it filled automatically as your library grows. ## Can I organize samples manually? Yes. Sample Vault never moves or renames your files. You can manually tag, categorize, and collect samples while AI tagging runs in parallel. ## What happens to my samples after downgrading my plan? Samples uploaded to cloud storage stay on our servers for 30 days after a downgrade. Re-upgrade within that window to keep access; after 30 days, anything over your new plan's cap is deleted. Local files on your disk are never touched. Cloud sync is separate and ends with the plan: the synced copy of your library's catalog is deleted when the subscription does, and browser access stops. Your library on your machine is untouched and stays fully searchable. If you subscribe again, turning cloud sync back on uploads it fresh from your machine. A failed payment doesn't trigger this. Sync keeps running while the payment is retried. ## Why is there a paid plan? Deep Tagging, the AI assistant, AI Collections, and cloud sync cost real money to run. The paid plans (Home Studio, Professional Suite) cover those costs. Local browsing, analysis, the everyday search, and playback are free on every plan and stay free. ## Is there a one-time purchase instead of a subscription? Yes. **Lifetime** is a one-time license that unlocks the desktop app for good, with all future updates included. AI features run through [your own key](/reference/plans-usage#bring-your-own-key-byok) rather than managed pools, and you can back up samples through your own [Dropbox or Google Drive](/reference/cloud-storage). It activates on more than one machine, and you can add a subscription later if you ever want managed AI and built-in cloud. See [Lifetime](/reference/plans-usage#lifetime) for the device limit and full breakdown. ## What audio formats does Sample Vault support? WAV, AIFF, MP3, FLAC, and OGG. ## Does Sample Vault integrate with my DAW? Yes. Drag samples directly into any DAW, or use [Ableton Link](/core-features/link-tempo-key#ableton-link) to sync sample playback with your DAW's transport. Combined with [Tempo and Key Lock](/core-features/link-tempo-key), you can audition samples in sync and in key with your project. Drag-to-DAW is a desktop-app feature — most DAWs reject drops from a browser tab. From the browser, right-click a sample → **Copy to clipboard** and paste into your DAW instead. ## Can I use my own Dropbox or Google Drive for cloud storage? Yes, on Professional Suite and Lifetime. Connect your own Dropbox or Google Drive under **Settings → Connections → Cloud storage** to route new sample uploads through your own account instead of Sample Vault's built-in cloud. Existing uploads stay where they are, and those uploads count toward your own provider's storage, not Sample Vault's caps. See [Cloud Storage](/reference/cloud-storage) for the full walkthrough. # Adding & Managing Samples Source: https://docs.samplevault.ai/get-started/adding-samples Learn how to import your sample library and understand the processing pipeline ## Overview Sample Directories is where you point Sample Vault at the folders on your disk that contain your audio files. The app scans recursively — pick your top-level samples folder and it handles every subfolder. You don't need to reorganize anything first; Sample Vault reads your library regardless of how it's structured. Sample Vault never deletes or moves your original files. They stay where they are on disk. ## What processing unlocks When you add a directory, Sample Vault starts working through it in the background. Rather than tracking internal stages, the **Library status** readout tells you what your library can *do* right now — each capability arrives on its own, and every sample becomes usable the moment the capability it needs is ready. | Capability | Cost | What you get | | - | - | - | | **Filter by brightness, width, and more sound characteristics** | Free | [Sound characteristics](/core-features/sound-profile) measured from the audio, waveform display, level-matched auditioning, and duplicate detection. Tempo and key filters already work from filenames; this fills in the files whose names said nothing. | | **Find similar, match a reference** | Free | [Similar Samples](/core-features/similar-samples) and reference-track matching, both working on how a sample sounds rather than what it's called. | | **Search by describing the sound** | Free | [Type what a sound is like](/core-features/search-modes) — `dark gritty kick`, `airy vintage pad` — and get it back even when the filename says nothing. [Deep Tagging](/core-features/sample-enrichment) sharpens these matches and makes genre, creator, and pack filters reliable. | | **Cloud backup** | Paid plans · storage cap | Samples backed up for access across devices. See [Plans & Usage](/reference/plans-usage). | Everything except cloud backup is free, unlimited, and automatic. [Deep Tagging](/core-features/sample-enrichment), which sharpens description matches, and [Deep Search](/core-features/search-modes), the optional AI pass over your whole library, both need a plan and an internet connection. You can start browsing and playing samples as soon as they're found, long before any of this finishes. ### When something can't be processed Files that can't be read — missing, moved, or in a format Sample Vault doesn't support — are counted separately in the Library status readout rather than holding your library permanently short of complete. Open a directory in Sample Directories to see what failed and why, grouped by cause, with the ones worth retrying separated from the ones that will always fail. ## Adding Your First Directory On first launch, setup looks for the folders your sample services already use — Splice, Loopcloud, ADSR, Noiiz — and offers the ones it finds, with the number of samples in each. Tick the ones you want and they're added to the library and start scanning; skip them and nothing is added. Custom locations are yours to add by hand. After that, you can add a directory from multiple places: * **File Browser**: Click the "+" button in the sidebar to add a directory directly from the file browser * **Settings**: Go to Settings → Sample Directories and click "Add Sample Directory" Select the folder containing your samples (this can be your main sample library folder) Choose processing options for the directory: * **Analysis speed**: Choose how much CPU and RAM to use — Battery, Balanced, Fast, or Max. The dialog shows the estimated resource cost for each level * **Per-folder upload and deep-tagging**: The dialog shows a tree of the folders being added with checkboxes for cloud upload and Deep Tagging on each one. Toggle them individually and watch the running totals — directory count, file sizes, and the number of samples that will be deep-tagged update as you go so you can see exactly what the scan will do before committing Your samples appear immediately in the Library once scanning completes. Processing continues in the background. ## Monitoring Progress Open **Library status** from the title bar. It shows how far along each capability is, how many samples can't be read, and an estimate of how much longer the rest will take. The readout also surfaces: * **Cloud Sync** state (paid plans) * A pause/resume control while processing is active, plus the Analysis Speed slider For the raw background queue — individual jobs, their errors, and retries — see the [Processing log](/reference/settings). You don't need to watch it, though. The first time a newly added library finishes analyzing, Sample Vault lets you know — in the app with suggested first searches to try, and with a system notification if you're in your DAW or another app at the time. If you're signed in, you'll also get an email when your library is ready, plus a reminder if the app was closed before analysis finished. ### When samples become available for each feature * **Browsing, playing, and search by name or token**: as soon as the file is found. Samples play from your disk, with no upload or tagging needed. * **Tempo and key filters**: as soon as the folder is scanned, for every file whose name says so. * **Sound Profile filters, waveforms, level-matched playback, and duplicate detection**: once the sound-characteristics row is ready. Tempo and key are filled in here for the files that named neither. * **Similar Samples and reference matching**: once *Find similar, match a reference* is ready. * **Search by description**: once *Search by describing the sound* is ready for that sample; until then it is still reachable by name and tokens. * **Genre, creator, and sample-pack filters**: available throughout, but only trustworthy for folders you've [deep-tagged](/core-features/sample-enrichment) — elsewhere the values are guessed from filenames. * **AI Assistant**: works from the moment a sample is found, and gets sharper as each capability lands. ## Managing Directories * **View Details**: Click on any directory to see detailed processing information and a list of all files * **Rescan Directory**: Use this when you've added new samples to an existing folder * **Retry Failed Files**: If some files failed processing, you can retry them individually or all at once * **Delete Directory**: Removes the directory from Sample Vault but doesn't delete your actual files ## Best Practices **Name your directories clearly**: The AI assistant can see directory names, so "Trap Samples 2024" is better than "New Folder (3)" * **Use nested folders**: You can add just your main samples folder - Sample Vault will process all subfolders * **Don't worry about file organization**: The AI will understand your samples regardless of how they're organized * **Add highest-priority folders first** so you can start working sooner * **Use meaningful folder and file names** (e.g., kick\_130bpm\_Cm.wav) to improve early filtering before AI tags are ready * **Let large libraries process overnight**: You can keep browsing and playing samples while analysis runs in the background ## Troubleshooting ### Why are my samples taking so long to process? Processing time depends on several factors: * **File size and number**: Larger collections take longer * **Audio analysis**: Each file needs tempo detection, key analysis, and audio fingerprinting * **Analysis speed**: The default Balanced preset leaves headroom for your DAW. Bump it up in your app **Preferences** (or in the System Status panel while analysis is running) to finish faster * **Deep Tagging queue**: Deep Tagging runs online, so it depends on the AI service being reachable Large libraries (10,000+ samples) can take several hours to fully analyze. The good news is you can browse and play samples immediately after scanning — analysis and deep-tagging continue in the background. The **Analysis Speed** slider in your app Preferences scales CPU and RAM usage together across four presets — Battery, Balanced, Fast, Max — so you can match it to whatever else you're doing on the machine. ### Some of my files show as 'Failed' - what happened? Common causes: * **Corrupted audio files**: The file may be damaged * **Unsupported format**: While Sample Vault supports most formats, some obscure codecs aren't supported * **Longer than ten minutes**: Anything over that is a recording rather than a sample. It isn't analysed for filters like tempo or key — it stays searchable by name and description, and its sound tags come from its opening seconds Right-click the failed sample or directory and use the "Analyze sample" option to retry analysis. ### My samples aren't showing up in search results Check that: * The capability your search needs is ready — check **Library status**, which names each one and how far along it is * You're searching in the right location (the Library vs AI Assistant) * Your search terms match your sample content * **AI Sample Suggestions** is enabled in preferences if you want suggested samples in your results # Sample Vault Documentation Source: https://docs.samplevault.ai/index Sample browser for music producers — local analysis, AI search, drag into any DAW. Desktop and browser on paid plans. ## Overview Sample browser for music producers. Local DSP analysis extracts tempo, key, and 16 sound characteristics from every file. AI tagging and Deep Search layer on top. Files stay where they are on disk. On Professional Suite, the same app runs in any browser at [samplevault.ai](https://samplevault.ai) — see [Browser Access](/reference/browser-access). ## Core Features Chat over your sample library. Describe what you need; get samples back. Filters, tags, collections, Sound Profile, and Map, Circle, and Scatter views. Drag straight into your DAW. Type a name or describe the sound — one search bar blends exact and sounds-like matches, with an AI pass on demand. Local analysis is free on every plan. Deep Tagging adds genres, tags, creator labels. Filter by tonality, brightness, attack, stereo width, and 12 more characteristics. An always-on-top search bar. Find and drag samples without leaving your DAW. Time-stretch and transpose in real-time. Sync to your DAW via Ableton Link. Arrange panels how you work. Save and switch layouts per session. ## Get Started Sample Vault for Windows, Mac (Apple Silicon), or Mac (Intel). No credit card required. Point Sample Vault at your sample folders. Files stay on disk; analysis runs in the background. Samples appear in the Library as soon as scanning completes. Search by name or by description. # Browser Access Source: https://docs.samplevault.ai/reference/browser-access Use Sample Vault from any browser tab — what works, what doesn't, and how to get in ## Overview Sample Vault runs in any modern browser as well as the desktop app. Sign in at [samplevault.ai](https://samplevault.ai) on any computer, click **Open app**, and the Library, Deep Search, Sound Profile, Similar Samples, and AI Assistant come up against your cloud library — same UI, no install. Browser access is included on **Professional Suite**. Demo Room, Home Studio, and Lifetime users see an upgrade prompt at sign-in. To try the experience without an account, open [samplevault.ai/demo](https://samplevault.ai/demo). ## What works in the browser The web tab is the same React app as the desktop window, so almost everything ports over: * **Library** — every filter, Sound Profile, tags, collections, sorting, the [extended search bar](/core-features/search-syntax), and the Map, Circle, and Scatter views * **[Deep Search](/core-features/search-modes)** — search by name or by describing the sound * **[Similar Samples](/core-features/similar-samples)** and the **[AI Assistant](/core-features/ai-assistant)** — Fast, Smart, and Creative chats plus workspaces and chat history * **Audio playback** with the [Waveform Visualizer](/core-features/waveform-visualizer) and [Tempo / Key Lock](/core-features/link-tempo-key) * **Library edits** — favorites, tags, collections, sample edits, and chat actions all sync back to your other devices ## What's desktop-only A few things need the OS or the local file system, so they only run in the desktop app. Each one shows a clear "desktop app required" prompt in the browser instead of failing silently: * **Library scanning** — adding directories, watching folders, rescanning, deleting samples from disk * **Local analysis** — BPM, key, and Sound Profile extraction * **Deep Tagging** — runs against samples on disk before they're uploaded * **Local-disk samples** — files that exist on your disk but haven't been uploaded to cloud storage are not visible in the browser library * **Ableton Link** — uses your local network * **Always-on-top pin**, **global shortcuts**, and the **[search overlay](/core-features/quick-actions)** — need OS window control ## How to tell you're in browser mode * A **globe pill** in the title bar marked "Browser mode". Hover it for the full list of what's available and what isn't. * The processing bar under the title bar reads "Analysis suspended — resume in the desktop app". * The Library status dialog opens with a one-line "X samples waiting — resume in the desktop app" verdict instead of a live processing view. ## Requirements * A current Chromium, Firefox, or Safari browser. * An active internet connection — every query goes to your cloud library; the browser tab can't work offline. * A **Professional Suite** subscription. The browser tab uses the same account as the desktop app — sign in with the email tied to your subscription. * **Cloud sync turned on** in the desktop app. Syncing is off until you turn it on in the app-data settings. The first sync uploads your library's catalog, which can take a few minutes on large libraries. ## Performance The web tab is slightly slower than desktop because every read is a round-trip to your cloud library, where desktop reads from local disk. The difference is most noticeable on filter-heavy panels and very large libraries. Search and Sound Profile filtering stay responsive at every library size. ## Troubleshooting ### "Subscription required" at sign-in Browser access needs a Professional Suite plan. Open the [pricing page](https://samplevault.ai/pricing) to upgrade, or keep using the desktop app on Demo Room, Home Studio, or Lifetime. ### The browser asks you to finish setup on desktop Your library reaches the browser through cloud sync, and syncing is off until you turn it on. Open the desktop app, enable cloud sync in the app-data settings, let the first sync finish, then reload the browser tab. If sync is already on, opening the desktop app once lets it complete the initial upload. ### Some samples are missing in the browser Only samples that have been uploaded to cloud storage are visible in the browser. Open the desktop app, right-click a directory, and choose **Upload samples to the cloud**. Missing files appear in the browser as soon as the upload finishes. See [Adding & Managing Samples](/get-started/adding-samples#the-processing-pipeline) for the full pipeline. ### Browser feels much slower than desktop A small slow-down is expected — every read goes over the network. If pages are taking several seconds, check your connection, then reload the tab. Heavy filter combinations on very large libraries (200k+ samples) can take a second or two to resolve; refining the query usually brings it back down. ### Drag-to-DAW doesn't work in the browser Most DAWs accept drops from the desktop app but not from a browser tab. If your DAW rejects the drag, right-click the sample and choose **Copy to clipboard** instead, then paste into the DAW. For full drag-to-DAW support, use the desktop app. # Cloud Storage Source: https://docs.samplevault.ai/reference/cloud-storage Use Sample Vault's built-in cloud or route uploads through your own Dropbox or Google Drive ## Overview When you enable cloud upload for a directory, Sample Vault pushes the audio files to a cloud storage provider so they're available on your other devices and in the [browser tab](/reference/browser-access). The default provider is Sample Vault's built-in cloud, which counts toward your plan's [storage cap](/reference/plans-usage#storage). On **Professional Suite** and **Lifetime**, you can also connect your own Dropbox or Google Drive account and route new uploads through that instead. Find it under **Settings → Connections → Cloud storage**. ## Built-in cloud Available on **Professional Suite**. No setup required — toggle **Upload to the cloud** per directory in the [Add Directory dialog](/get-started/adding-samples#adding-your-first-directory), or right-click a folder → **Upload samples to the cloud** later. Files count toward your plan's [storage cap](/reference/plans-usage#storage). ## Connect Dropbox Scroll to the **Cloud storage** section and click **Connect Dropbox**. Dropbox opens in a new tab. Sign in and grant Sample Vault permission to write files into a dedicated `/Apps/Sample Vault` folder inside your Dropbox. The Cloud storage panel switches to showing your account label and used / available figures within a few seconds. Once connected, new sample uploads go through Dropbox automatically. Existing uploads stay where they already are. ### How files are organized in your Dropbox * New uploads land at `/Apps/Sample Vault/`, mirroring the folder structure inside the source sample directory. * Sample Vault doesn't read anything from your Dropbox outside the `/Apps/Sample Vault` folder. * Sample Vault doesn't keep its own copy of the audio — it forwards the bytes to Dropbox during upload and serves the file back through a short-lived signed URL when playback needs it. Renaming or moving files inside the Dropbox web UI breaks the link to that sample inside Sample Vault until the next reconcile. Use Sample Vault to manage your samples; treat the `/Apps/Sample Vault` folder as storage, not as a place to reorganize. ### Disconnect Click **Disconnect** in the Cloud storage panel to stop syncing. Sample Vault revokes its access token. Files already in `/Apps/Sample Vault` are not deleted — you can keep them as a backup or remove them manually from Dropbox. ## Connect Google Drive Scroll to the **Cloud storage** section and click **Connect Google Drive**. Google opens in a new tab. Sign in and grant Sample Vault permission to create and access files in a dedicated `Sample Vault` folder inside your Drive. Sample Vault only sees files it created — never the rest of your Drive. The Cloud storage panel switches to showing your account email and used / available figures within a few seconds. Once connected, new sample uploads go through Google Drive automatically. Existing uploads stay where they already are. ### How files are organized in your Drive * New uploads land in a top-level `Sample Vault/` folder in your Drive, with subfolders mirroring the structure inside the source sample directory. * Sample Vault uses the `drive.file` OAuth scope, so it can only see files it created. It cannot read anything else in your Drive. * Sample Vault doesn't keep its own copy of the audio. During playback the bytes pass through a short-lived Sample Vault URL that's authenticated with your stored Google credentials — your files are never made publicly shareable in Drive. Renaming or moving files inside the Google Drive web UI doesn't break the link (Drive uses stable file IDs), but trashing or deleting them does. Treat the `Sample Vault/` folder as storage, not as a place to reorganize. ### Disconnect Click **Disconnect** in the Cloud storage panel to stop syncing. Sample Vault revokes its OAuth token with Google. Files already in `Sample Vault/` are not deleted — you can keep them as a backup or remove them manually from Drive. ## Switching providers Pick a default upload destination right from **Settings → Connections → Cloud storage** once at least one provider is connected. The **Default upload destination** card lets you toggle between Sample Vault storage, your Dropbox, and your Google Drive without disconnecting any of them. New uploads use whichever destination is currently selected; samples already in the other providers stay where they are and remain playable. * **Sample Vault storage** is only selectable on plans that include built-in cloud — Professional Suite. Otherwise the option is shown but locked behind an upgrade prompt. * **Your Dropbox** and **Your Google Drive** are only selectable while that provider is connected. The card disables the option (with the existing connection card still visible) if the provider isn't linked yet. ## Storage caps and quotas | Provider | Cap | | - | - | | Built-in cloud (Professional Suite) | 200 GB, counts toward your plan | | Dropbox | Your Dropbox plan's storage. Sample Vault shows live used / available figures in the panel | | Google Drive | Your Drive plan's storage. Sample Vault shows live used / available figures in the panel. Workspace accounts with unlimited storage show "—" for the cap | When a quota is full, new uploads pause. Existing uploaded samples remain accessible. ## Browser access Dropbox- and Drive-stored samples are playable in the [browser tab](/reference/browser-access) the same way built-in-cloud samples are. Dropbox streams through short-lived Dropbox URLs; Drive streams through short-lived Sample Vault URLs that proxy the bytes from your Drive — your provider's bandwidth covers playback in both cases. ## FAQ ### Will connecting a BYO provider migrate my existing cloud samples? No. The existing uploads stay where they already are and remain playable. Only new uploads after connecting are routed through the newly chosen provider. ### Can I connect more than one provider at once? Yes. You can keep Dropbox, Google Drive, and the built-in cloud all linked at the same time and switch which one new uploads go to from the **Default upload destination** card in **Settings → Connections → Cloud storage**. Only one is active at a time; the others stay linked so previously uploaded samples keep playing. ### What happens if I disconnect a BYO provider? Sample Vault stops uploading new files to that provider and revokes its access token. Samples already stored there aren't deleted — they're just no longer reachable from Sample Vault until you reconnect. Files uploaded to the built-in cloud or to a still-connected provider keep working. ### Is the cloud storage the same as the cloud sync database? No. Your library catalog (sample metadata, tags, favorites, chat history, search index) syncs through a separate path that always goes through Sample Vault. Cloud storage providers only handle the audio files. ### Why does Google Drive playback go through a Sample Vault URL instead of a direct Drive link? Privacy. Google Drive doesn't offer a Dropbox-style "give me a short-lived public URL" endpoint. The only way to get a public link is to mark the file as shareable with anyone who has the URL — which would mean every Sample Vault user's audio files are technically public. Instead, Sample Vault mints a short-lived signed URL pointing at its own server, which proxies the bytes from your Drive using the OAuth token you granted. Your files stay private; Sample Vault never holds them. ### Can I connect a BYO provider on Lifetime? Yes. Uploads to your own Dropbox or Google Drive don't require a subscription, so Lifetime holders can connect a BYO provider, set it as the default destination, and upload to it. The subscription requirement applies only to **built-in cloud** (Sample Vault's own storage); BYO files live in your own account against your own provider quota. Cross-device playback through samplevault.ai also works on BYO, as long as the connection stays linked. Cloud upload isn't part of Demo Room or Home Studio: connecting a provider is available on Professional Suite and Lifetime. # Keyboard Shortcuts Source: https://docs.samplevault.ai/reference/keyboard-shortcuts Sample Vault keyboard shortcuts ## Overview ## Global Shortcut The **Open Assistant** shortcut works even when Sample Vault isn't the active window — useful for jumping from your DAW into the assistant without switching windows manually. The shortcut saves automatically. ## Application Shortcuts These work when Sample Vault is the active window, as long as no text input is focused. | Action | Shortcut | | - | - | | Play previous | `1` / `↑` | | Play / pause | `2` / `Space` | | Play next | `3` / `↓` | | Resume from pause | `Shift + Space` | | Skip back / forward | `Shift + ←` / `Shift + →` | | Copy playing sample to clipboard | `Cmd/Ctrl + C` | | Toggle selection | `Cmd/Ctrl + click` | | Range-select | `Shift + click` | | Select all loaded | `Cmd/Ctrl + A` | | Focus library search | `Cmd/Ctrl + F` | | Reveal the assistant bar | `Cmd/Ctrl + I` | | Help & feedback | `Cmd/Ctrl + Shift + /` | The copy shortcut only fires when no text is highlighted — otherwise the standard text copy behaviour applies. Skip back and forward move the playing sample by a few seconds a press, and repeat while held, so you can travel a long way into a multi-minute recording. They apply to whatever is playing, wherever it is playing from. Because they take `Shift`, they stay clear of the Map view's arrow-key navigation below. The library search shortcut works from anywhere — even while another input is focused — and opens a Library panel first if none is open. The assistant-bar shortcut also works from anywhere and is rebindable under **Settings → Keyboard Shortcuts**. Audio shortcuts are automatically disabled while you're typing in a search or chat input. Selection shortcuts work in every list of samples — the Library, [Deep Search](/core-features/search-modes), [File Browser](/core-features/file-browser), [Similar Samples](/core-features/similar-samples), and [AI Assistant](/core-features/ai-assistant) chat chips. See [Multi-select & batch actions](/core-features/sample-browser#multi-select-batch-actions). ## Section Shortcuts Shortcuts that only fire inside specific parts of the app. ### Library search input | Action | Shortcut | | - | - | | Move through suggestions | `↑` / `↓` | | Apply highlighted suggestion | `Tab` | | Run search or apply highlighted suggestion | `Enter` | | Deep Search your whole library | `Cmd/Ctrl + Enter` | | Close suggestions, then unfocus input | `Esc` | ### Search overlay Summon the [search overlay](/core-features/quick-actions) with its configurable global shortcut (default `Cmd/Ctrl + Shift + Space`), even when Sample Vault isn't focused. Once it's up, these keys act on it: | Action | Shortcut | | - | - | | Move through results | `↑` / `↓` | | Play or pause the highlighted result | `Space` | | Copy the highlighted result, dismiss the overlay | `Enter` | | Deep Search, or ask a question | `Cmd/Ctrl + Enter` | | Add a filter | `F` | | Apply suggested filters | `A` | | Clear all filters | `C` | | Grab the last 15 seconds into Tape | `G` | | Open or close Tape | `T` | | Open or close sync settings | `S` | | Open the Sample Vault window | `M` | | Step back, then dismiss the overlay | `Esc` | Arrow keys move focus out of the search field and onto the result they play, so `Space` and `Enter` act on that sample and `Esc` puts you back in the field. `Space` and the single-letter keys fire only when the search field isn't focused, so they never collide with typing a query. ### Library filters With a Library panel active and no input focused: | Action | Shortcut | | - | - | | Open filters | `F` | | Apply suggested filters | `A` | | Save the current filters | `Cmd/Ctrl + S` | | Clear all filters | `C` | | Deep Search your whole library | `D` | ### Sample Map When the Library panel is showing the Map view and is the active panel, the arrow keys jump to the nearest visible sample in that direction and play it. The viewport recentres on the new sample if it would otherwise fall off-screen. | Action | Shortcut | | - | - | | Navigate to nearest sample | `←` / `→` / `↑` / `↓` | | Open sample context menu | Right-click a dot | ## Troubleshooting ### The global shortcut isn't working 1. Check no other application is using the same combination 2. Use 2–3 keys (e.g. `Ctrl+Shift+S`) — single-key shortcuts conflict with system input 3. Avoid common system shortcuts (`Ctrl+C`, `Ctrl+V`, etc.) 4. Restart Sample Vault after setting a new shortcut # Plans & Usage Source: https://docs.samplevault.ai/reference/plans-usage Understand subscription plans, usage limits, and billing information ## Overview For pricing and feature comparisons, visit the [pricing page](https://samplevault.ai/pricing). ## Plans Sample Vault has four plans: * **Demo Room** (free) — the full local app, free forever, with a weekly allowance to try the AI features * **Home Studio** (subscription) — for individual producers * **Professional Suite** (subscription) — for power users who want cloud storage, sync, and unlimited AI * **Lifetime** (one-time purchase) — own the desktop app outright instead of subscribing; see [Lifetime](#lifetime) ## How AI Usage Works Library size is unlimited on every plan — your samples stay on disk, and analysis (BPM, key, sound characteristics) is always free and runs locally. Only AI operations are metered, and what meters them depends on the operation and your plan. **Deep Tagging** — the pass that gives every sample its real genre, creator, and sample pack — runs across your whole library on every paid plan, with no cap on size or count. On Demo Room you can deep-tag one folder to try it. **Deep Search**, the assistant, and the kit designer run under fair use on Professional Suite, so a normal production day never runs out. On Home Studio the assistant carries a monthly allowance while Deep Search and the kit designer run under fair use. **AI Collections** and **library auto-organization** are paid-plan features that run under fair use, with no allowance. ### Starter Pools Every account without an active subscription — Demo Room, and Lifetime holders who haven't added a subscription — can try the AI features for free: * A weekly allowance of **AI actions** shared across the assistant's AI features — [Deep Search](/core-features/search-modes), the assistant chat, the [Sketch Designer](/core-features/sketch), and library [Suggestions](/core-features/suggestions) all draw from it. Every action costs the same, whichever feature it came from. Your first allowance is larger than the weekly one, so there's room to try each feature properly before it settles into the weekly rhythm. The allowance refills each week; unused actions don't carry over. * **Deep Tagging on one folder**, so you can see your real genres, creators, and packs before deciding. On a paid plan these AI features run under fair use (with a monthly allowance for the Home Studio assistant), and Deep Tagging covers your whole library. When an allowance runs out, Sample Vault shows what you used and the date it refills, so you can wait for the next one or move to a paid plan. Lifetime holders can instead route AI through their own key. ## Bring Your Own Key (BYOK) Route AI requests through your own OpenRouter API key instead of Sample Vault's managed pools. BYOK is available on **Professional Suite** and **Lifetime**. Two modes are available: * **Key only** — every AI call uses your key; no Sample Vault pool is touched. * **Pools first, then key** — operations draw from your plan's pools, and silently fall through to your key once a pool is empty. On **Lifetime**, your own key is how the AI features run — there are no managed pools to fall back on. ### Checking that a key works A key OpenRouter accepts is not always a key that can do the work. Your OpenRouter account's privacy, guardrail, and provider settings decide which providers a request may reach, and your credit balance decides whether it runs at all. A spending limit is not a balance: OpenRouter won't charge your card unless you pre-pay or turn on auto top-up. Sample Vault chooses its own model for each feature, so an account restricted to one model family can block work even when that family is allowed. Deep Tagging, search, the assistant, and collections don't all run on the same model. Sample Vault can check a saved key for you. It confirms the key is recognised, reports the remaining balance, and tries a minimal request against each model it relies on, naming the reason for anything that comes back blocked. The check runs when you save a key and can be run again at any time from your API key settings. ## Assistant Memory The assistant can remember your preferences, the conventions you work by, and what you're currently working on, and carry that context between sessions. It's part of **Professional Suite**, and available on **Lifetime** when you bring your own AI key. **Home Studio** and **Demo Room** don't carry memory between sessions. Everything the assistant remembers is visible and editable — see [AI Assistant → Memory](/core-features/ai-assistant#memory). ## Lifetime **Lifetime** is a one-time license that unlocks the desktop app for good, with all future updates included. There's no recurring subscription. What it includes: * **The full desktop app** — unlimited library size, the Library and all its views, Sound Profile, Similar Samples, Quick Actions, tempo and key lock, Ableton Link, and unlimited projects. * **AI through your own key.** Lifetime doesn't come with managed AI pools, so [Deep Search](/core-features/search-modes), Deep Tagging, the assistant, and AI collections run on a [key you bring yourself](#bring-your-own-key-byok). Without a key, you still get the same free [weekly AI allowance](#starter-pools) as Demo Room. * **Your own cloud, optional.** There's no built-in cloud storage, but you can back up and sync through your own [Dropbox or Google Drive](/reference/cloud-storage) — no subscription required. * **Activated on your devices** — see [Device activation](#device-activation) below. You can add a Home Studio or Professional Suite subscription on top of Lifetime whenever you want managed AI without your own key; Professional Suite also adds built-in cloud storage and sync. While the subscription is active its allowances apply; your Lifetime license stays yours regardless. ### Device activation A Lifetime license is tied to your devices rather than to an active subscription. The first time you open the app on a machine, give the device a name and activate it. The license can stay active on **up to three devices** at once. To move the license to a different machine once you've reached the limit, revoke one of your activated devices from the **Manage devices** page, then activate the new one. Revoking frees the slot right away. ## Storage Storage caps apply only to samples uploaded to cloud storage. Local files on your disk never count. | Plan | Cloud storage | | - | - | | Demo Room | 0 GB (local only) | | Home Studio | 0 GB (local only) | | Professional Suite | 200 GB | | Lifetime | 0 GB (local only) | Built-in cloud storage and sync come with **Professional Suite**. Storage caps are max-preserving across concurrent entitlements (the highest one wins) and never shrink on downgrade. On **Professional Suite** and **Lifetime**, you can also route uploads through your own [Dropbox or Google Drive](/reference/cloud-storage) instead of the built-in cloud. Those uploads count toward your own provider's storage, not Sample Vault's caps. Lifetime has no built-in cloud, so this is its backup and sync path. ## Browser Access [Browser access](/reference/browser-access) — using Sample Vault from any modern browser at [samplevault.ai](https://samplevault.ai) — is included on **Professional Suite**, since it works against the samples you've uploaded to cloud storage. Demo Room, Home Studio, and Lifetime users see an upgrade prompt at sign-in. ## Monitoring Usage Open **Settings → Billing & Usage** for your allowance meters. The avatar in the top navigation also shows a small dot when one is running low — open the menu to see which. ## When You Hit Limits * **Demo Room Deep Tagging trial used** — Once you've deep-tagged your one free folder, upgrade to a paid plan to deep-tag your whole library, or route Deep Tagging through your own key on Lifetime. * **Weekly AI allowance used (Demo Room)** — The shared AI features pause until the allowance refills the following week. Active features like [Deep Search](/core-features/search-modes), the assistant chat, and the [Sketch Designer](/core-features/sketch) stop running; passive assists like library [Suggestions](/core-features/suggestions) and prompt ideas simply go quiet. Upgrade to a paid plan (or, on Lifetime, add your own key) to keep going without waiting. Local browsing, playback, and the everyday free Library search keep working as normal. * **Home Studio assistant allowance used** — The assistant pauses for the rest of the month; Deep Search, the kit designer, and Deep Tagging keep running under fair use. Upgrade to Professional Suite for an assistant with no monthly allowance. * **Fair-use cap hit (paid)** — You'll see a brief slow-down message; capacity recovers within seconds to a minute. * **Storage cap hit** — New uploads pause; existing uploaded samples remain accessible. ## Managing Your Subscription * **Demo Room users**: Click "Upgrade Plan" in **Settings → Billing & Usage** to see paid options. * **Paid subscribers**: Use "Manage Your Subscription" to update payment details, change plans, or cancel. # Settings Source: https://docs.samplevault.ai/reference/settings Customize Sample Vault preferences and account settings ## Overview ## Assistant ### AI Assistant Instructions A persistent prompt the AI reads on every conversation. Use it to describe your music style, sample library, and how you want the assistant to behave. **What to include:** * Your music style: "I mainly produce dark techno and industrial music" * Your workflow: "I prefer one-shots over loops" or "I like to layer multiple percussion samples" * Your sample library: "My 'Dark Ambient' folder contains atmospheric textures, while 'Drums' has acoustic and electronic percussion" * How you want the assistant to behave: "Be concise in your responses" or "Always suggest 3-5 samples when I ask for recommendations" **Example instructions:** ``` I'm a techno producer focused on underground, industrial sounds. I have separate folders for: - Kicks (organized by sub-genre: Acid House, Hard Techno, Minimal) - Percussion (acoustic drums, industrial hits, found sounds) - Synths (bass patches, lead synths, pads) When I ask for samples, prioritize darker, grittier sounds over clean commercial samples. I prefer building tracks from individual elements rather than using full loops. ``` The AI assistant reads these instructions with every conversation, so the more specific you are about your music style and preferences, the better recommendations you'll get. Think about your typical music-making session. If you always start by looking for kicks, and you always prefer dark, industrial sounds, telling the AI assistant this information means it will immediately understand what you're looking for instead of giving you generic suggestions. ### Assistant Memory Manage what the assistant remembers between sessions, alongside your assistant instructions. See [AI Assistant](/core-features/ai-assistant#memory) for what it remembers and how it learns. * See every entry the assistant has saved, and edit or delete any of it * Switch individual entries off without deleting them, or turn memory off entirely * Confirm or dismiss entries flagged as possibly out of date Entries stay in your library's database on your machine. Memory is part of Professional Suite, or Lifetime with your own AI key. ## Preferences ### System Settings **Autostart on System Startup**: When enabled, Sample Vault launches automatically when your computer starts up. This is handy if you use Sample Vault daily - your samples will already be deep-tagged and ready when you sit down to make music. New installs turn this on by default; you can switch it off here at any time. **Search overlay visibility**: Controls when the [search overlay](/core-features/quick-actions) appears — whenever Sample Vault isn't the focused window, always, or off. Also toggled from the system tray menu. ### Processing Speed A single slider with four presets — **Battery**, **Balanced**, **Fast**, **Max** — controlling how much of your machine Sample Vault uses while it works through your library. A live readout shows the estimated memory and CPU cost of the preset you've picked, so you can weigh speed against whatever else you're running. * **Battery** — minimal resource usage, ideal when your DAW is open * **Balanced** — good speed without competing for resources * **Fast / Max** — work through large libraries quickly when your machine is free Whichever preset you choose, processing runs at a lower system priority than your other apps, so it gives way to your DAW and stays out of the way of live playback. The same slider appears in the **Add Directory** dialog and in the **System Status** panel while processing is running. ### Display Settings **Theme**: Choose the colour theme used across the app. Your choice is saved and applied every time Sample Vault opens. Each theme sets not just colours but its own corner radius — from sharp and utilitarian to softly rounded. Themes are grouped into Neutral, Accessibility, and Studio classics. Neutral — daily drivers, darkest to lightest: * **Midnight** — the deep-navy dark; the default * **Graphite** — a clean neutral-charcoal dark * **True Black** — pure black, easy on OLED screens * **Dim** — a softer charcoal grey, easier on the eyes during long sessions * **Ember** — a warm, low-blue-light dark for late nights * **Paper** — a warm off-white light * **Daylight** — a clean, bright light theme for daytime rooms Accessibility — maximum contrast for low-vision use or bright environments: * **Contrast** — black and white with a high-visibility yellow accent * **Contrast Light** — white and black with a high-visibility blue accent Studio classics — one bold accent each: * **Wavetable** — near-black with a neon-green accent * **808** — warm black with a burnt-orange accent * **Session** — neutral grey with an amber accent * **Console** — graphite with an electric-blue accent * **Pocket** — a bone-white light theme with a punchy red accent * **Screen** — dark olive with an acid-green accent **Font**: Pick the typeface used throughout the interface. **UI Scale**: Scale the entire interface from 50% to 200% to suit your display. ### Playback Adjustments applied while you audition samples. Your files on disk are never modified. * **Default volume** — the level used for auditioning across the app * **Normalize playback loudness** — match perceived loudness across samples, with an adjustable target peak * **Trim leading / trailing silence** — start playback at the audio and stop at its end * **Remove DC offset** — strip constant DC bias without shifting transient timing * **Boundary declick** — optional fades at the sample start and end, including silence-trimmed edges; turn it off to preserve the original attack and tail. Switching, seeking, and pause/resume use separate short smoothing that does not follow this fade length. * **Render samples as heard when dragging or copying** — bake the active adjustments into the file your DAW receives, instead of sending the untouched original ### Privacy & Data Turn usage analytics on or off, and choose whether to receive product emails. Marketing email preferences need a signed-in account. ## Sample Libraries Manage which public sample collections appear in your library. ### Public Sample Libraries Sample Vault includes curated public sample collections. You can enable or disable individual collections to keep your library focused on what you actually use. * Go to **Settings → Sample Libraries** * Toggle collections on or off * Disabled collections are hidden from the Library and search results If your library feels cluttered with samples you don't recognize, check Sample Libraries — you may have public collections enabled that don't match your workflow. ## Organization Sample Vault keeps your library organized as it grows — correcting and discovering categories by how samples sound, tidying the genre and tag lists, and building collections. **Organization** is where you shape what it does and manage the vocabulary it works with. Processing Speed and the label lists that used to live under a separate Processing tab now sit in **Preferences** and here, respectively. ### Automatic organization Turn the background organization on or off, and choose which passes run on their own. A **house rules** note — written in plain language, like "keep liquid and neurofunk separate" or "no tempo-based collections" — is followed across every pass. You can also run a full pass over your whole library on demand instead of waiting for the background chain. Because it keeps organizing on its own in between, an on-demand run has a short cooldown afterwards. It's available on paid plans — see [Deep Tagging](/core-features/sample-enrichment) for what the pass does and [Plans & Usage](/reference/plans-usage) for how it's metered. ### Your taxonomy Categories, genres, tags, creators, and sample packs are all managed here. Edit them, add your own, and add categories that the sound-based pass is allowed to assign. * **Lock** a genre or tag so tidying never merges, renames, or prunes it. Anything you create or edit by hand is protected already * Renaming, moving, or deleting an entry asks for confirmation and tells you how many samples the change will affect, so a bulk edit is never a surprise ### Recent activity Every automatic pass and on-demand run is logged with what actually changed — the categories it named, the genres and tags it tidied, and the collections it built. ## Connections The **Connections** section (formerly **API Keys**) covers external services Sample Vault talks to on your behalf. ### Bring Your Own Key (OpenRouter) Route AI calls through your own OpenRouter API key instead of Sample Vault's managed pools — available on **Professional Suite** and **Lifetime**. Two modes — **Key only** uses your key for everything; **Pools first, then key** falls through to your key only when your plan's pools are exhausted. See [Plans & Usage → BYOK](/reference/plans-usage#bring-your-own-key-byok). ### Cloud storage Connect a Dropbox account to route new sample uploads through your own Dropbox instead of Sample Vault's built-in cloud. The panel shows account label and used / available storage once connected. See [Cloud Storage](/reference/cloud-storage) for the full setup, file layout, and disconnect behavior. ## App Data The **App Data** section provides access to application data directories and maintenance tools. * **Reveal directories**: Direct links to open cache, logs, and other app-related directories on your system * **Cloud sync** (Professional Suite): Turn syncing of your library's catalog (tags, analysis results, and edits) to your account on or off. It's off until you turn it on; enabling it is what makes your library available in the browser and on other machines. Audio files are not part of it and stay on your disk by default: [upload](/reference/cloud-storage) the ones you want to reach from anywhere. Turning it off deletes the synced copy from your account and leaves the library on your machine untouched and fully searchable; turning it back on later uploads it again from that machine. The app restarts when you change it. * **Database backups**: Create and restore backups of your local database * **Rebuild File Tree**: Regenerate the virtual file tree if directories appear incorrect * **Repair Database**: Run an integrity check on your local database and fix what it finds — rebuilding indexes that have drifted out of sync, and rebuilding the database file itself if its internal structure is damaged. Use it when scans, deletes, or batch operations stall unexpectedly, or if a backup is refused because the database looks corrupted. Safe to run any time — no data loss. * **Compact Database**: Reclaim disk space the database is still holding. Removing folders or samples doesn't shrink the database file on its own, so it can stay much larger than the library it now describes. Compacting gives that space back. Worth running after you remove a large folder. It rewrites the whole file, so it takes a while on a big library, needs free disk space roughly equal to the current database size, and pauses background processing while it runs. Your samples aren't touched. Database backups are experimental and may not handle very large datasets yet. ## Processing log The Processing log is the raw record of background work: scans, analysis, deep-tag runs, uploads, and their errors and retries. It opens on failures, because that is almost always why you're here. For how far along your library is overall, use **Library status** instead. Each entry shows a timestamp, the sample or directory involved, and the outcome, so you can find why a particular sample never processed. A failed entry carries the error the job returned. Open it to read the message in full, or copy it to paste into a bug report or a Discord question. Two actions sit above the list: * **Retry failed** requeues the failed jobs matching your current filters, with a fresh attempt count. Samples that are missing or in an unsupported format will fail the same way again — review those in [Sample Directories](/get-started/adding-samples), which groups failures by cause. * **Clear log** removes completed entries. Failed entries are kept unless you explicitly opt in, since their error is the only record of why a sample didn't process. Pending and processing jobs are never affected. ## Account Your account settings cover basic profile information and session management. ### Account Information Shows your registered email address - this is where you'll receive important notifications about processing status, subscription changes, and feature updates. ### Getting Started Checklist Brings the Getting Started checklist back into the sidebar, including any steps you set aside. Steps you've already completed stay ticked off, since the app works those out from what you've actually done rather than from the checklist itself. ### Sign Out Signs you out. You'll need to sign back in to access your samples and preferences. # Troubleshooting Source: https://docs.samplevault.ai/reference/troubleshooting Installation, offline mode, and support for Sample Vault ## Overview For problems specific to a feature, see that feature's page — most have a Troubleshooting section at the bottom. This page covers installation, offline mode, and support. ## Where to get help Click the **Help & feedback** icon in the sidebar (or press `Ctrl/Cmd+Shift+/`) to open one dialog with everything: documentation, changelog, keyboard shortcuts, bug reports, feature requests, free-form messages, and the Discord invite. Bug reports automatically attach app version, OS, recent errors, and up to the last 10 MB of the local log file — you can turn that off before submitting. If the app never gets past its startup screen, that screen offers the same bug report, so a failure to start can be sent without reaching the sidebar. Reporting works even when the local library is the thing that failed. ## Installation & Updates ### A new version hasn't reached you yet New releases roll out to existing installs in stages, so a version can be announced a few days before your copy offers it. Nothing is wrong, and you don't need to reinstall. If you'd rather not wait, download the current version from the [downloads page](https://samplevault.ai/downloads) and install it over your existing copy. Your library, settings and analysis stay as they are. ### v2.0.0 Migration v2.0.0 introduced a new database architecture that is not compatible with previous versions. After updating, you'll be prompted to migrate. We recommend skipping the migration and starting fresh with a reanalysis of your samples. A fresh install from the [downloads page](https://samplevault.ai/downloads) is only required if the migration does not complete successfully. ### macOS Apple Silicon Build Redownload the app from the website and select the correct version for your CPU architecture. A previous bug caused Apple Silicon users to receive Intel builds. ## Offline Mode After signing in once while online, all local features work offline: * Sample browsing, filtering, and search by name and tokens * Audio playback with Tempo and Key Lock * Sound Profile filtering * Similar Samples These features require an internet connection: * [AI Assistant](/core-features/ai-assistant) chat * [Deep Search](/core-features/search-modes) — the AI pass that matches samples by your description * [Deep Tagging](/core-features/sample-enrichment) * Cloud sync and backup ## Samples won't play or show up as missing If files you added before suddenly won't play, show as missing, or can't be written to, Sample Vault has most likely lost read access to where they live on disk. The app only sees a sample if the account running it can reach that path. This is most common with samples kept on a network share or external drive: * The drive or share isn't mounted, or mounted at a different path than when the samples were added. * On a machine with more than one user account, the share is mounted under a *different* account than the one running Sample Vault, so the app inherits permissions that can't read or write those files. Make sure the location is connected and readable by the same account you run Sample Vault under, at the same path it had when you added the samples. Remounting it correctly restores playback and clears the missing state with no reanalysis needed. If a sample has genuinely moved, playing it offers a **Locate** option that points it at its new path and brings along other files that moved with it. If you reorganised a whole folder — sorted sounds into new subfolders, renamed things — each sample folder has a **Reconnect moved files** action that finds every sample's file wherever it ended up and reattaches it, keeping the tags, descriptions, favourites, and collections already on it. Files are matched on name and size, and on their contents where that isn't enough to tell two apart; anything it can't identify with confidence is left alone rather than guessed at. Nothing on disk is touched. Run this *before* scanning the folder for new samples — a scan treats a moved file as a brand-new one and adds an unprocessed copy alongside the original. Reconnected samples keep their existing place in the folder tree until the next full scan. If files were deleted for good and you want them out of the library, each sample folder has a **Remove missing files** action that checks every sample and removes the ones whose files no longer exist, instead of resolving them one at a time. It never touches files on disk, keeps cloud-backed samples, and — so a disconnected drive can't wipe a folder — removes nothing when the folder itself is offline. In that case, reconnect the drive or share first. ## Database is misbehaving If scans, deletes, or batch operations stall unexpectedly — or a backup is refused because the database looks corrupted — run **Settings → App Data → [Repair Database](/reference/settings#app-data)**. It runs an integrity check on the local database and fixes what it finds, rebuilding drifted indexes and, if the file's internal structure is damaged, rebuilding the database file itself. Safe to run any time — no data loss. ## Database is larger than my library The database file doesn't shrink on its own when you remove folders or samples, so it can stay much bigger than the library it now describes. Run **[Compact Database](/reference/settings#app-data)** to give that space back. It rewrites the whole file, so allow some time on a large library and leave free disk space roughly equal to the current database size. ## Browser tab [Browser access](/reference/browser-access) needs a Professional Suite subscription. If you see a "Subscription required" screen at sign-in, open the [pricing page](https://samplevault.ai/pricing) to upgrade, or keep using the desktop app. Only samples uploaded to cloud storage appear in the browser library. If samples are missing in the browser but visible in desktop, open the desktop app and right-click the directory → **Upload samples to the cloud**. They appear in the browser as soon as the upload finishes. Library scanning, sample analysis, Deep Tagging, Ableton Link, and drag-from-local-disk are desktop-only — the browser surfaces a "desktop app required" prompt for each. # User Interface Reference Source: https://docs.samplevault.ai/reference/ui-reference Navigation bar and sample context menu ## Overview ## Navigation Bar The navigation bar at the top of the app provides access to all major features. ### Main Navigation * **AI Assistant** — chat-based sample discovery and production help * **Library** — your full collection with search, filters, Sound Profile, and the Map, Circle, and Scatter views * **Settings** — directories, processing, AI, billing, account ### Title Bar Controls * **Project Selector** — switch between projects; the active project's BPM and key are applied to playback automatically * **Directory Drawer** — folder icon opens the [File Browser](/core-features/file-browser) with your configured sample directories * **System Status** — opens the **Library status** readout showing how far along each capability is, how many samples could not be read, cloud sync state, and a pause/resume control with the Analysis Speed slider while processing is active. See [Adding & Managing Samples](/get-started/adding-samples) for what each capability unlocks * **Subscription Info** — current plan and a low-pool indicator. Click to open Billing & Usage * **Help & feedback** — single dialog for docs, changelog, shortcuts, bug reports, feature requests, and Discord (`Ctrl/Cmd+Shift+/`) ### Playback Controls * **Link** — enables [Ableton Link](/core-features/link-tempo-key#ableton-link) and auto-enables Tempo Lock * **Tempo Lock** — time-stretches every sample with detected tempo to a global BPM * **Key Lock** — transposes every sample with a detected key to a target key See [Link, Tempo and Key Lock](/core-features/link-tempo-key) for details. ### Window Controls * **Pin Window** — keeps Sample Vault on top of other applications. The navigation bar becomes more compact when pinned * **Drag handle** — when pinned, drag the grip in the center of the navigation bar to move the window * **Resize** — works like any other window, pinned or not ## Sample Context Menu Right-click any sample anywhere in the app for the full action set. ### Sample management * **Sample Info** — full metadata, tags, technical details * **Edit Sample** — modify metadata (only on samples you uploaded or as admin) * **Quick property editors** — change **Key**, **Scale**, **Tempo**, **Creator**, **Sample pack**, **Type** (loop/one-shot), or **Category** without opening the full editor. Creator and Sample pack open a searchable picker; Tempo suggests your library's most common BPMs * **Tags / Genres** — add/remove dialogs that mirror the Sample Info picker. Right-click any chip in the Filters dialog or Sample Info pane to **Rename** or **Delete** it across your whole library * **Classify sample** — re-run AI audio tagging without redoing local analysis * **Add/Remove from Favorites** * **Copy to Clipboard** — copy the file for pasting into your DAW * **Open File Location** — jump to the sample's folder on disk or in the Library ### AI actions * **Discovery** — Find similar samples, What genre is this, Suggest complementary samples * **Creative** — Create a beat using this, Suggest arrangement ideas, Help me layer this * **Analysis** — Analyze the musical characteristics, What key is this in, Describe the mood ### Collections * **Add/Remove from Collection** — drop into custom collections