Missed the latest GitHub check; these facts may be older.
Overview
AudioMuse-AI analyzes the audio in your music library on your own server and uses the results for clustering, similarity playlists, a 2D music map, and text, lyrics and 20-second recording search. It connects to Navidrome, Jellyfin, LMS, Lyrion, Emby and Plex, including several at once with cross-server duplicate detection. It runs as Flask and worker containers on PostgreSQL 15, or as native packages for macOS, Linux and Windows.
Who it is for: Self-hosters with a Navidrome, Jellyfin or Plex music library
Strengths
Supports Navidrome, Jellyfin, LMS, Lyrion, Emby and Plex, several servers in one deployment
Analysis runs locally without metadata or external APIs; no ads or paywalls
Native packages bundle embedded PostgreSQL, web UI and workers; no Docker needed
Swagger API docs at /apidocs/ and AMD64/ARM64 images, with NVIDIA GPU variants
Weaknesses
Requires PostgreSQL v15; other versions may cause errors
Needs AVX2 CPU; VMs must pass through host CPU, noavx2 images are experimental
Native apps are unsigned; macOS build is Apple Silicon only, Windows x86_64 only
Major or minor releases can require re-running the full library analysis
What it needs
RAM ≥ 8 GB
GPU optional
Docker
Needs PostgreSQL 15, Navidrome, Jellyfin, LMS, Lyrion, Emby or Plex server