Self-Hosted Music Streaming with Smart Discovery and Social Features
Navidrome and Jellyfin let you stream your own music library from a home server with privacy and FLAC quality. But the biggest thing missing is music discovery: no 'you might also like' recommendations, no radio stations, no friend activity, no shared playlists. Spotify's algorithm is the primary reason people stay despite privacy concerns. Users want to own their music AND discover new music without feeding their listening habits to a corporation.
Don't build another music server. Build a recommendation plugin for Navidrome. Use MusicBrainz metadata + Last.fm's public artist similarity data + local audio feature extraction (librosa or Essentia) to generate 'Daily Mix' style playlists from the user's own library. Ship as a Navidrome companion service that runs alongside it. The Subsonic API gives you access to the library. The recommendation engine is the product. If it works, Navidrome might integrate it upstream.
landscape (4 existing solutions)
Self-hosted music streaming solved the playback and library management problem (Navidrome is excellent). But it completely ignores discovery, which is Spotify's real product. The missing piece is a recommendation engine that runs locally: analyze your library's audio features and metadata, suggest similar artists from MusicBrainz/Last.fm data, and generate smart playlists without sending listening habits to any server. Open-source audio fingerprinting and local ML models make this technically feasible in 2026.