"Play mix" in the discover UI used to spread new songs through the whole
library playlist, so the share of new songs depended on how many there
were. It now plays the new songs with songs the user has played mixed in
evenly at the rule's familiar_ratio (30% by default), the same ratio that
sets how much of the Subsonic playlist is familiar.
Which songs in each mix playlist were played recently is now recorded
when the mix is built (cluster_mix_playlist.known_song_ids) and returned
with the mix; the UI marks them as "Played". Mixes built before this fall
back to using the library songs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018opPJMJjLWr4RqdG6vuK6p
Each mix rebuild now also finds songs that aren't in the library: the
user's Listenbrainz recommendations they've never played, plus an LB
Radio prompt per mix when a Listenbrainz token is configured. Each
candidate goes to the mix whose genre profile it fits best and is
skipped if it's already in the library or was played recently.
discover/ is a new React app to listen to the mixes: library songs
stream from Subsonic and suggested songs are found on YouTube with
yt-dlp, both proxied through playlistd with Range support for seeking.
Suggestions can be liked (kept in a "Liked new songs" list) or hidden.
- discover: candidate gathering, genre fit scoring, library filtering
- web: /api/discover routes, Subsonic/YouTube streaming proxy
- youtube: yt-dlp search (skipping hour-long results) and audio URLs
- mix tags now hold library genre names, for display in the UI
- config: listenbrainz.token, discover.ui, discover.ytdlp, discover.format
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018opPJMJjLWr4RqdG6vuK6p
A new cluster_mix_rule clusters a Listenbrainz user's listens from the
last few months by MusicBrainz genre (with tracks played in the same
session pulled together) and maintains one Subsonic playlist per cluster,
filled with the cluster's most played tracks plus unheard library songs
from the best matching library genres.
- listenbrainz: fetch raw listens since a timestamp and bulk recording
genres, handling rate limits
- cluster: tag TF-IDF + co-listen smoothing, spherical k-means with k
chosen by silhouette score
- mix: match cluster profiles to library genres and pick songs, with a
per-artist cap; reuse playlists across rebuilds by genre overlap
- subsonic: getGenres, getSongsByGenre, rename/delete playlist, genre and
MBID fields on tracks, API error detection
- daemon, migration, POST/GET /api/playlist/mix and README docs
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018opPJMJjLWr4RqdG6vuK6p