Research of the optimal wavelet selection on entropy function

Abstract

The selection of the appropriate wavelet is the key issue in the wavelet packet de-noising. In this paper, it takes the entropy function as the evaluation criteria for the best packet basis function. The entropy is calculated by the coefficients of wavelet packet decomposition of speech signal to determine the appropriate decomposition. At the same time, four kinds of packet basis will be used to denoise in computer simulation experiments with wavelet packet threshold algorithm. The simulation results show that the optimal wavelet bases, which should be selected by two entropy function, not only can eliminate background noise to a large extent, but also raise the SNR of voice signal. © 2012 Springer-Verlag.