Denoising algorithm based on wavelet adaptive threshold

Abstract

the main idea of denoising algorithm based on wavelet adaptive threshold is that speech signals should be packet transformed to get the wavelet coefficients used in optimal wavelet. Since the signal and the noise have different relevance, there will be different attenuations in wavelet decomposition process. Based on above characteristics, the appropriate threshold can be calculated by a new threshold function and the minimum mean square algorithm, even if the noise coefficients can be removed and the signal coefficients can be saved. Finally, the retained coefficients can be reconstructed to restore the original signal for the purpose of de-noising.