Post

AudioMuse-AI

AudioMuse-AI is a self-hosted tool that uses sonic analysis to rediscover forgotten songs and generate groove-aware playlists for your music library, without external APIs. It integrates with Jellyfin, Navidrome, Emby, Lyrion or Plex and runs a Flask web UI backed by Redis Queue workers and PostgreSQL.

AudioMuse-AI
In Development
This script is currently in active development and may be unstable or incomplete. Use in production environments is not recommended.

Installation

Default install:

1
bash -c "$(curl -fsSL https://raw.githubusercontent.com/community-scripts/ProxmoxVED/main/ct/audiomuse-ai.sh)"
CPU: 4 cores RAM: 8192 MB Disk: 20 GB OS: Debian 13

Default Credentials

UsernamePassword
adminNone

Configuration

Config file:

1
/opt/audiomuse-ai_data/audiomuse.env

Notes

The generated web UI password is saved to /root/audiomuse-ai.creds inside the container (user: admin). Open http://[IP]:8000 to log in.
Connect your music server before running an analysis: edit /opt/audiomuse-ai_data/audiomuse.env and set MEDIASERVER_TYPE (jellyfin, navidrome, emby, lyrion, plex) plus the matching URL and credentials, then 'systemctl restart audiomuse-ai audiomuse-ai-worker audiomuse-ai-worker-high'.
The installer downloads roughly 2.2 GB of ML models (musicnn, CLAP, Whisper, silero, gte) into /opt/audiomuse-ai_data/model, so the setup takes a while. Configuration, models and caches live in /opt/audiomuse-ai_data so they survive updates; the PostgreSQL database holds the analysis results. Sonic analysis itself is CPU- and memory-intensive.
Running two analysis workers can use significant RAM. If the container runs out of memory, disable audiomuse-ai-worker-high or increase the assigned RAM.

Web Interface

Port: 8000

Source code
Loading...