Open-source AI DJ: LLMs plus user data could beat Spotify recommendations
Reddit user proposes LLM trained on listening data and text for perfect playlists...
Get AI news that actually matters
One email a day. Zero fluff. Join 10,000+ professionals.
Deep Dive
A Reddit user asked for a way to generate playlists from a song or list of songs, suggesting that combining user listening data with text data like reviews and social media could work with an LLM. They noted that current tools like Last.fm APIs are limited and that Spotify's recommendations are best for them but locked away. The user speculated that bulk listening data might be hidden behind proprietary platforms.
Key Points
- Proposes LLM trained on listening metrics + text data (reviews, social media) for playlist generation
- Current open-source options (Last.fm API, Navidrome) are limited or 'janky'
- Main obstacle: bulk listening data is locked inside Spotify, YouTube Music, etc.
Why It Matters
An open-source AI DJ could break Spotify's monopoly on music discovery and empower users with transparent recommendations.