DeepCrate

Appearance

Using Your Own Organizer ​

DeepCrate can be aware of what's already in your library, for duplicate detection in the discovery queue without organizing your music itself. This is the right setup if you already rely on a tool like wrtag, beets, Lidarr, or a slskd post-processing script to move and tag your downloads.

The Two Features Are Independent ​

FeatureWhat it doesTouches filesystem?
library_duplicateReads your library from a Subsonic-compatible server and flags discoveries you already ownNo, API only
library_organizeMoves completed slskd downloads into a library folderYes, moves/copies/deletes files

If you let another tool organize, you want library_duplicate enabled and library_organize disabled (which is the default). DeepCrate will never read or write files on your music library when configured this way.

Example Configuration ​

yaml
catalog_discovery:
  subsonic:
    host: "http://navidrome:4533"
    username: "your_subsonic_username"
    password: "your_subsonic_password"

library_duplicate:
  enabled: true
  auto_reject: false   # set true to silently drop duplicates instead of flagging them

library_organize:
  enabled: false       # leave off, your existing tool handles layout

The Subsonic block under catalog_discovery is shared with the Catalog Discovery feature; you don't need a separate connection.

What DeepCrate Reads ​

DeepCrate calls the Subsonic getAlbumList2 endpoint and caches (artist, album, year) tuples in its own SQLite database. It does not:

  • read your music files
  • read your music directory
  • write anything back to the Subsonic server
  • care about how your files are laid out on disk

Any Subsonic-compatible server works: Navidrome, Gonic, Airsonic, Airsonic-Advanced, etc. Your organizer's job is to keep that server's index up to date; DeepCrate's job is to read it.

How Matching Works ​

Duplicates are matched at the album level on normalized (artist, album) pairs. Normalization is lowercase + trim, there is no diacritic folding, no leading-The stripping, and no track-level matching. See Limitations below.

Sync Schedule ​

The library is re-synced periodically by the library-sync background job:

SettingDefaultDescription
LIBRARY_SYNC_INTERVAL (env var)86400 (24h)Seconds between automatic syncs

After each sync, all pending queue items are re-checked against the new library snapshot. If auto_reject: true, items now found in the library are moved to rejected; otherwise they're flagged with an "In Library" badge.

Verifying It's Working ​

Once you've enabled library_duplicate and pointed catalog_discovery.subsonic at your server:

  1. Open the Queue page. Near the top you should see Library: N albums · synced X ago. If N is 0 or the timestamp never updates, the sync isn't reaching your server, check the server logs for Subsonic errors.
  2. Click Sync Now on the Queue page to trigger a sync immediately instead of waiting for the schedule.
  3. Use the Hide Owned toggle in the queue filters to confirm duplicates are being detected.

You can also trigger a sync via the API:

bash
curl -X POST http://localhost:8080/api/v1/jobs/library-sync/trigger \
  -H "Authorization: Basic <credentials>"

Limitations ​

  • Matching is naive. Because normalization is only lowercase + trim, slight differences will not be treated as duplicates:
    • Blue Öyster Cult in your library vs. Blue Oyster Cult in the queue
    • The Chemical Brothers vs. Chemical Brothers
    • Siamese Dream (Deluxe Edition) vs. Siamese Dream
  • Album-level only. DeepCrate has no awareness of individual tracks within an album. If a recommendation is for a single track, it's resolved up to its parent album (when mode: "album") before being matched.
  • No filesystem fallback. If you don't run a Subsonic server, there is currently no way to make DeepCrate aware of your library.
  • Sync is best-effort. If your Subsonic server is unreachable when the job runs, the existing cached snapshot is preserved and the next run will retry.