Spotify Audio Features and Exercise Tolerance in Cycling: An Exploratory Analysis

The authors explored how audio features from self-selected music may be associated with exercise tolerance in endurance cycling, measured through time-to-exhaustion (TTE) trials. Participants (N = 29) privately selected high-tempo (120–140 BPM) tracks on Spotify, and then completed two counterbalanced TTE trials at ~80% of their individual Peak Power Output (Wmax), one with music and one in no-music. On average, participants cycled 5.10 minutes longer with music compared to no-music (d = 0.54). Nine Spotify audio features (valence, energy, tempo, loudness, speechiness, instrumentalness, liveness, acousticness, and danceability) were extracted for each track via the Spotify API and averaged across each participants' playlist. Compared to general music, cycling tracks were characterized by higher liveness, energy, loudness, and tempo, but lower danceability, instrumentalness, and speechiness. K-means clustering identified three clusters of self-selected music, although differences in ΔTTE across clusters did not reach statistical significance. The largest estimated gains were observed for the high valence-moderate energy cluster (M = +8.04 min). This work provides preliminary evidence that music with elevated valence may be associated with exercise tolerance.