Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.patchline.ai/llms.txt

Use this file to discover all available pages before exploring further.

Patchline builds a user-scoped search view of your catalog so you can search by free-text query, mood, genre, BPM range, ISRC, or sonic similarity. This powers Aria’s catalog tools and the dashboard search bar.   All tiers can search. Sonic-similarity ranking improves with catalog size and runs across the full catalog regardless of tier.

​
Web dashboard

Sidebar → Catalog → search bar at top. Type any query: 
"dreampop 76 bpm"
"tracks like 'Sleeper'"
"F minor 4/4"
"sad and slow"
"ISRC GBUM71029604"
Results rank by combined relevance: keyword match + sonic similarity + recency.

​
Via Aria

​
Via MCP

catalog_search is the semantic search — pass any text query, get ranked results. browse_catalog is the listing endpoint — paginate through your full catalog without a query.

​
What you can search by

Query typeExampleWhat it returns
Title"Sleeper"Tracks with that title
Artist"Mira"Tracks by that artist
ISRCGBUM71029604The single matching track
Genre / mood tags"dreampop", "melancholy"Tracks tagged with those
BPM range"between 100 and 120"Tracks in that range
Key"F minor", "D major"Tracks in that key
Time signature"3/4", "7/8"Tracks in that meter
Energy / valence"high energy", "sad"Tracks above/below thresholds
Sonic similarity"like [track ID]"Tracks with cosine-similar audio embeddings
Composite"dreampop 76 bpm female vocal"All conditions AND’d
The semantic layer means natural language works — “slow sad ballads for late night” returns the right tracks even if those exact words aren’t in any tag.

​
How search works

  • Catalog metadata is normalized on every catalog write.
  • Sonic features from sonic analysis ride alongside title, artist, ISRC, release date, and platform metadata.
  • Similarity features are available for tracks with completed audio analysis.
  • Results are scoped to your workspace and ranked by field match, freshness, audio similarity, and requested filters.
When you query, Patchline returns relevance-ranked results and merges in fresh presentation metadata before rendering.

​
Examples

​
Find a specific track by ISRC

You: GBUM71029604
Aria: That's "Bohemian Rhapsody" by Queen. Already in your catalog
      (imported 2026-04-12). Want full intel or just the audio features?

​
Find tracks for a playlist

You: I'm building a chill playlist — find me 10 tracks in my catalog
     that fit
Aria: 10 best fits (low energy + acoustic-leaning + slow tempo):

  1. Sleeper (Mira) — 76 BPM, F# minor, energy 0.31, acousticness 0.78
  2. Long Way (Tessa) — 82 BPM, A minor, energy 0.34
  3. Velvet (Velvet Bird) — 71 BPM, D minor, energy 0.28
  4. ...

  Want me to make a smart link with these 10?

​
Find similar tracks

You: Find tracks in my catalog that sound like 'Late Bloom'
Aria: Top 5 by sonic similarity (cosine 0.7+):

  1. Late Bloom (the query track itself, 1.00)
  2. Bright Drop (Cleo Bell) — 0.81
  3. Soft Rain (Tessa) — 0.79
  4. ...

​
Browse without a query

You: List my catalog
Aria: 47 tracks total. Showing first 10:
  [paginated list with title, artist, BPM, key, last activity]

  Want to filter? "all dreampop", "tracks added this month", etc.

​
Limitations

  • Search is only as good as the indexed data. Tracks without completed sonic analysis won’t match audio-feature queries.
  • Embeddings are catalog-only. “Find similar tracks” works within your catalog. For cross-catalog similarity, see Scout agent.
  • Index lag is ~30 seconds. A freshly imported track is searchable within ~30 seconds of import completing.

​
Pricing

ActionCredit cost
Catalog search query0 (cached)
catalog_search MCP call0
browse_catalog MCP call0
Search is effectively free. Only the Aria chat conversation wrapping the search counts as 1 credit.

​
FAQ

No. Search is user-scoped. Team and label workspaces can search across the shared workspace catalog, but not other customers’ catalogs.
Search can have a ~30 second lag after import. Wait a moment and refresh. If still missing after 5 minutes, the import may have failed — check the catalog UI for an error state.
Patchline compares audio-similarity features from completed sonic analysis and filters out weak matches. Metadata, genre, and audience context can also affect the ranking.
Saved searches / smart playlists are on the roadmap. v1 you re-run the query.