Name a song, then decide how much each trait counts. Nine audio features — danceability, energy, mood, and the rest — each on a weight slider. Drag one up and the ranking re-sorts live: you author what “similar” means, the machine measures the distance. Matched against a …-track corpus.