Music Recommendation Engine
Overview
A music recommendation engine that extracts audio features from songs (like tempo, timbre, and energy) and uses Approximate Nearest Neighbor search to quickly find and recommend the most similar tracks.
The Challenge
Efficiently searching high-dimensional song feature vectors at scale, balancing recommendation accuracy against query speed as the song library grows.
What I Learned
Learned how to extract and normalize audio feature vectors, build and tune ANN indexes, and evaluate recommendation quality against exact nearest-neighbor baselines.
Tech Stack
PythonANNFlask
Key Features
- Audio Feature Extraction
- Approximate Nearest Neighbor Search
- Fast Similar-Song Recommendations