How Spotify predicts what you want to listen to

Spotify uses a combination of collaborative filtering, audio analysis, and exploration-exploitation strategies to power its recommendation engine. Collaborative filtering finds users with similar listening habits and cross-recommends music. After acquiring Echo Nest in 2014, Spotify also analyzes raw audio waveforms converted into spectrograms and processed through convolutional neural networks to extract features like tempo, energy, and danceability — enabling recommendations even for brand-new tracks with no listeners. The system also deliberately surfaces songs users might not like to refine taste profiles over time.