Turning 10 years of Spotify history into playlists with Claude
Someone used a 10-year Spotify listening history export as raw material for Claude to build custom playlists. The Spotify export includes skip data, so Claude could tell which tracks were rarely listened to all the way through. Since LLMs struggle to read efficiently, the first step was having Claude clean up the export into an 'LLM-friendly' format, producing CSV and files, plus a table of most-listened artists.
This cleanup step didn't need a top-tier model. Next, working inside Claude Code, the cleaned data was analyzed for fun insights like preferred genres, characteristics of favorite music, and listening eras over time, and a website was built to visualize the results.
Key points
- Used a 10-year Spotify listening history export as input for Claude to build custom playlists
- Skip data in the export let Claude identify tracks that were rarely fully played
- was first ed into CSV/ files that are easier for an LLM to read
- This cleanup step worked fine on a lighter model, saving cost
- A website was built afterward to visualize the listening data