MetaBrainz as an organisation has never much dabbled in (artificial) intelligence, but a number of recent factors have led to the team doing some exciting behind-the-scenes work over the last few months.
Lately more and more potential contributors have come to MeB interested in working on AI projects, and with ListenBrainz we have an excellent dataset. With a current focus on playtesting and finetuning our playlist features we also have the perfect use-case.
So, without further ado, we invite you to test the beta version of our new AI-powered playlist generator:
AI Brainz Playlist Generator (beta)
(Extra bonus: the AIBrainz server is also running a beta preview version of our upcoming UI redesign! Future UI updates will be published on this beta server until we release the new design.)
Unlike most ListenBrainz tools, you do not need an LB account to test this feature. However, you must give the tool greater than usual permissions to access your listening history on your device. This trade-off allows us to get more eyes and data into the system. How we tackle this may be revisited in future.
For our more technical users, we collect user data from music streaming services, social media, and wearable devices, as long as they are linked to your device/s. This data includes user listening history, playlist data, song metadata, and user demographics. The data is then preprocessed, including data cleaning, feature extraction, and normalization. Next, the system uses a clustering algorithm (hierarchical clustering) to group similar songs and artists together. The system uses some dimensionality reduction techniques (principal component analysis), to reduce the dimensionality of the data and identify underlying music listening patterns.
AI Brainz (AIB) then uses the collected data to try and generate a playlist for you while attempting to side-step the classical cold-start problems inherent in many recommendations systems. We hope that using a quick and simple data analysis and a lot of heavy AI lifting we can generate a playlist to your tastes in just a few seconds.
And please leave feedback for us in the comments, or via your preferred method of communication (for example, but not limited to, forums, twitter, carrier pigeon…). Have fun!