NeptuneHub/AudioMuse-AI — reverse-engineered prompt

Build me a self hosted app called AudioMuse AI for people who want to rediscover songs already in their own music library. It should connect to servers like Jellyfin, Navidrome, LMS, Lyrion, and Emby, scan and analyze the actual audio, and then help people explore their collection without depending on tags or outside music services.

I want a simple web app with an easy first run setup wizard, login support, and a dashboard where someone can run an initial analysis and then use features like smart playlists from a vibe or text prompt, similar songs from one track, a visual music map, song paths between two tracks, sonic fingerprint playlists from listening habits, song alchemy with add and subtract choices, plus text and lyrics search. It should let people export playlists back to their music server.

Please make it runnable locally with Docker, and keep it practical for home server use. If anything is unclear, check the project docs online and follow the current setup described there.