> ## Documentation Index
> Fetch the complete documentation index at: https://docs.samplevault.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# 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** — stays visible in the sample browser, search, and similar-sample results. The rest are hidden. Hidden copies are never deleted; your files on disk are left exactly as they are.

<Note>
  Duplicate detection works on samples that have been analysed. Older samples
  join in as their directories are re-scanned.
</Note>

## 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 again everywhere.
* **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.

<Tip>
  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.
</Tip>