Join my party! How can we enhance social interactions in music streaming?
summary
Collaborative web app maps songs onto a 2‑D mood plane using valence and arousal coordinates from Last.fm, with 10,000 tracks streamed as 15s Deezer previews. New uploads get feature vectors (onset rate, spectral flux, entropy, autocorrelation), a random forest selects the most informative features, and nearest‑neighbor averaging assigns mood coordinates. Group voting moves a cursor that picks and an automated decision‑tree mixer blends tracks by tempo/key/beat regularity; private parties and bot votes keep playback running.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:04Music selection interfaces
- 3:17Social voting system
- 5:09Private party mode
- 7:59Mood prediction algorithm
- 9:23Automatic song transitions
- 11:09Demo and resources
- 11:30Q&A
