Matched wavelets for musical signal processing using evolutionary algorithms

The non-stationary nature of musical signals presents challenges for conventional signal analysis methods. Wavelet transforms offer a powerful tool for capturing both temporal and frequency information simultaneously. This study introduces a novel approach to enhance wavelet analysis in music processing by utilizing matched wavelets optimized through evolutionary algorithms, specifically tailored for musical signals within the context of Indian Classical Music (ICM). Various evolutionary algorithms including Particle Swarm Optimization (PSO), Genetic Algorithm (GA) and Differential Evolution (DE) were investigated. The proposed method optimizes wavelet parameters to match the characteristics of a given signal resulting in a customized CWT filter bank. The scalogram accurately highlights the fundamental frequency and its harmonic components. The efficacy of this approach is validated through comparisons with established techniques such as Short-Time Fourier Transform (STFT) and S-Transform. The designed wavelets achieve a high correlation coefficient in signal reconstruction, outperforming standard continuous wavelets. The customized wavelets not only facilitate the detailed analysis of signal components but also ensure robust signal reconstruction. The use of matched wavelets in feature extraction has shown promising results in tasks such as swara recognition and instrument identification in monophonic music.