No tracks yet -- let's fix that!

Search for songs you love and add them here. We'll analyze their audio DNA to find your next favorites.

Find Your First Track

How Recommendations Work

NextTrack uses content-based filtering combined with external data enhancements. Each track has audio features (energy, valence, danceability, acousticness, tempo) forming a 5-dimensional vector. We compute the centroid of your playlist to represent your taste.

External data actively boosts recommendations:

Artist metadata from MusicBrainz (country, type, decade) enables filtering. Your feedback (likes/dislikes) teaches categorical preferences for future recommendations.

No account required - playlist stored in browser session (7 days). All external data cached in Redis (24-hour TTL) - no database storage.