lyrid
A music universe: a canonical sky of artists and genres you explore through real listening.
lyrid turns the music world into a night sky. Every artist is a star, every genre a nebula, similarity forms the routes between them — one canonical map for everyone, with a personal fog of war over it. You light up your own sky by actually listening: previews are scans, full listens are landings, and your scrobbles fuel the journey.
Named after the Lyrids — the meteor shower radiating from Lyra, the lyre of Orpheus: music falling from the sky as stars.
Two modes
- Exploration — fog of war, a starship with fuel, light as currency, quests and hidden treasures. Chosen once, played for years.
- Creative — the whole sky open from the first minute: a daily instrument for studying music.
Built on open data
The universe is assembled locally from open dumps and datasets — MusicBrainz, ListenBrainz, Discogs, Wikidata, Wikipedia, AcousticBrainz — with similarity and the sky layout computed by lyrid itself. Listening goes through official previews and embeds; scrobbling connects via ListenBrainz.
Status
Pre-alpha. The stars are in and the routes between them: a MusicBrainz full export builds the canon (~3M artists with countries, years, release groups and URL relationships), and the ListenBrainz relations dataset adds ~7M similarity edges with the brightness derived from them. The sky layout and the renderer come next. Watch this repository.
Development
Requires Rust (see rust-version in Cargo.toml), Node LTS with pnpm, and Docker for the development database.
# Fill the universe from a MusicBrainz full export (~7 GB, downloaded once):
# Add the similarity graph (~117 MB, CC0):
&& && && &&
Documentation
lacodda.github.io/lyrid — guides, reference, and the architecture decision records.