As I understand it, Sonic Fingerprint currently reads play history directly from the connected media server (Plex/Jellyfin/Navidrome/etc). This breaks down for anyone whose actual listening happens through a client that doesn't report plays back to that media server — for example, listening to a Plex library exclusively via Home Assistant's Music Assistant, which doesn't write play events back to Plex at all. In that setup, Sonic Fingerprint only ever sees a small, arbitrary slice of real listening, or none at all.
Many users in this position already scrobble everything accurately to Last.fm or ListenBrainz via their playback client. It would be great if AudioMuse-AI could optionally ingest listen history from one of these sources instead of (or in addition to) the connected media server, for use in Sonic Fingerprint. A few possible shapes this could take, roughly in order of implementation effort:
`. A simple authenticated ingest endpoint (tokenized URL) that any external tool could POST play events to, decoupled from any specific service
2. Native pull-based support for the Last.fm and/or ListenBrainz user APIs, similar to how existing media server adapters work
Support for standard scrobble webhooks
Any of these would let Sonic Fingerprint reflect real listening habits for users whose primary playback path doesn't route plays back through their media server. Happy to help test if useful — thanks for the great project!
P.S. - I bet the folks at Home Assistant would be happy to assist if so needed.
As I understand it, Sonic Fingerprint currently reads play history directly from the connected media server (Plex/Jellyfin/Navidrome/etc). This breaks down for anyone whose actual listening happens through a client that doesn't report plays back to that media server — for example, listening to a Plex library exclusively via Home Assistant's Music Assistant, which doesn't write play events back to Plex at all. In that setup, Sonic Fingerprint only ever sees a small, arbitrary slice of real listening, or none at all.
Many users in this position already scrobble everything accurately to Last.fm or ListenBrainz via their playback client. It would be great if AudioMuse-AI could optionally ingest listen history from one of these sources instead of (or in addition to) the connected media server, for use in Sonic Fingerprint. A few possible shapes this could take, roughly in order of implementation effort:
`. A simple authenticated ingest endpoint (tokenized URL) that any external tool could POST play events to, decoupled from any specific service
2. Native pull-based support for the Last.fm and/or ListenBrainz user APIs, similar to how existing media server adapters work
Support for standard scrobble webhooks
Any of these would let Sonic Fingerprint reflect real listening habits for users whose primary playback path doesn't route plays back through their media server. Happy to help test if useful — thanks for the great project!
P.S. - I bet the folks at Home Assistant would be happy to assist if so needed.