NeptuneHub/AudioMuse-AI v0.7.1-beta

AudioMuse AI v0.7.1-beta: Song Alchemy

on GitHub

Release Date: October 18, 2025

This is a great release for self-hosted music lovers, with AudioMuse-AI v0.7.1-beta we’re introducing Song Alchemy, a new way to explore and blend musical moods. You can now ask to hear something like song A and song B, but different from song C. For example, find tracks with the dreamy vibe of indie pop and the energy of synthwave, but without the heavy electronic feel. AudioMuse-AI computes a balance between these moods to surface songs that match your unique taste.

This release also adds more control over your analysis workflow. You can now choose which music libraries to analyze using the MUSIC_LIBRARIES environment variable. Leave it empty to scan all libraries, or specify a comma-separated list of names or paths. For Lyrion, use folder paths like /music/myfolder, while Jellyfin and Navidrome accept library names directly.

Highlights

• Feature Expansion: Song Alchemy

  • Added new Song Alchemy feature with visualization support.
  • Fixed UI button styles and improved smartphone responsiveness.
  • Removed unnecessary ten-song filter for more flexible analysis.

• Mood Similarity Enhancements

  • Added MOOD_SIMILARITY_THRESHOLD environment variable.
  • Enabled mood similarity by default in similarity.html.
  • Improved clustering (duplicate removal, starvation prevention).

• Performance & Compatibility Updates

  • Introduced multithreading for Voyager index processing.
  • Updated Dockerfile for improved CPU compatibility.

• Music Library & Analysis Improvements

  • Added MUSIC_LIBRARIES env variable to target specific libraries (#106).
  • Updated artist field logic with fallbacks and improved track analysis (#70, #103).

• Documentation & Maintenance

  • Multiple README.md and hardware.md updates.
  • Removed unused files and improved playlist fingerprint handling (Jellyfin).

What's Changed

• Devel -> Main by @NeptuneHub in #113

Full Changelog: v0.7.0-beta...v0.7.1-beta