Getting started

Audience: First-time users setting up a local library. Goal: Move from checkout to a searchable, analyzed SQLite database. Type: tutorial index

The first useful result is not a fully automatic DJ set. It is a searchable map of your own music and a short list of candidates you would probably not find by browsing folders alone.

The shortest path is: install, scan, open the UI, analyze a small batch, search, then decide by ear. Your audio stays where it is. Scan and analysis add information to local SQLite databases; they do not reorganize or rewrite the source files.

What each step gives you

1. Scan turns a folder tree into a library you can browse and filter.
2. Analysis adds the evidence needed for sound-based search and set previews.
3. Search reduces a large library to a listening shortlist.
4. SET or crate building turns useful candidates into an editable working list.
5. Export creates an M3U or CSV file for the next part of your workflow.

Pages

• Quickstart: the shortest path through scan, serve, and first analysis.
• Install: prerequisites, package extras, FFmpeg, frontend, and docs build notes.
• First library: how scanning works and what the database stores.
• First analysis: choose analysis by the result you want, then configure limits and devices.

What you need first

• A local folder of audio files.
• A local SQLite path where the app can create or open the library.
• FFmpeg available on PATH or through DJ_TRACK_SIMILARITY_FFMPEG.
• Optional model dependencies if you want SONARA, MAEST, MERT, MuQ, or CLAP analysis.

Privacy habit

Treat the SQLite database, logs, reports, generated indexes, and classifier artifacts as private library data. They can include paths, tags, model scores, and listening notes.