Comparative performance study between Spectral Subtraction and Discreet Wavelet Transform for Speech enhancement

摘要

In this paper we present a comparative performance study between Spectral Subtraction (SS) and Discreet Wavelet Transform (DWT) methods for Speech enhancement. We have implemented three extended approached based on SS and two other ones based on the DWT. We have used a speech corpus extracted from NOISZEUS database [5]. Two different noises, Babble and Street, have been used to validate the implemented methods. Experimental results show that the DWT based approaches outperformed the classical methods.