← All projects
Shazam Clone
Functional · group project Audio · DSP 2025Audio fingerprinting that identifies songs from a library of 1000+ tracks.
- Python
- librosa
- NumPy
- SciPy
- scikit-learn
- Streamlit
- sounddevice
Overview
A fully functional Shazam-style song identifier built as a group project. It fingerprints audio by extracting constellation peaks from FFT spectrograms, hashes them, and matches a microphone recording against a database of over a thousand songs using time-offset histogram voting.
Highlights
- FFT spectrogram peak (constellation) fingerprinting
- Hash-based lookup with time-offset voting
- Identifies from a library of 1000+ songs
- Live microphone capture and matching
- Streamlit tools for fingerprint inspection