Musical Feature Engineering with Wavelet Analysis for Music Recommendation

This project presents a novel method for extracting musical features via signal processing and wavelet analysis. Utilizing the human intuition of constructive sound wave interaction, we construct the classical techniques of signal processing in Fourier analysis. Building upon Fourier analysis, this project utilizes wavelet transforms to extract individual beats, and contextualizes them within larger rhythms for comparison and recommendation. Fundamental limitations of time- and frequency-localized information are explored in a signal processing context, with emphasis and interpretation on the human experience of music consumption