Tired of your music recommendations? Support undiscovered artists and find new tracks for your playlists, expertly handpicked by BBC Introducing. No algorithms, no genres and no personalisation β listen and tune into what sounds good to you now. Powered by humans, not AI.
Hey Mathieu,
How do you determine which tracks have a "similar vibe" without relying on genres or other metadata?
Do you use any audio analysis techniques to group similar-sounding tracks together? I'd love to learn more about the technology behind it.
Congrats on the launch!
@kyrylosilin Hey! We've got a blog post coming up but in short we use some specialised (and open source) machine learning models to extract 8 different indicators (danceability, if it's more relaxed or more party oriented, if it's sad or happy sounding, if it's electronic or acoustic and if it's instrumental only, etc.) and we use dimensionality reduction techniques to compress the 8 dimensions into 2. Can update here with the blog post when it's ready :)
Congrats to the BBC Orbit team on the launch! This is an exciting way to discover fresh, undiscovered talent without relying on algorithms. Are there any features for creating custom playlists from the tracks you discover?
The absence of genres and personal data makes it feel like a genuine music exploration journey. Plus, the limit of five reveals a day adds a layer of intentionality, encouraging users to savor each track. Itβs a fantastic way to discover new music while supporting upcoming artists.
BBC Orbitβs focus on music discovery without relying on genres or algorithms is refreshing . It allows listeners to tune into what truly resonates with them in the moment.
Telebugs
BBC Orbit
Diaflow
TabsMagic
BBC Orbit
Premarket Bell