Music Instrument Classification Using Generalized Gaussian Density Modeling

Abstract

In this work, subband coefficients obtained from one dimensional wavelet decomposition of isolated note samples of different instruments has been modeled using generalized Gaussian density. By using only model parameters, the classification of music instruments has been done by calculating the Kullback-Leibler divergence between two different densities. The effect of different mother wavelet functions used in wavelet decomposition has also been investigated.