A Denoising Method of Speech Signal Based on Hilbert-Huang Transform

Abstract

HilbertHuang Transform(HHT) is a new and effective method of analyzing nonlinear and nonstationary time series.In this paper,a denoising method of speech signal based on HHT is presented.First,speech signal polluted by white noise is decomposed into several Intrinsic Mode Functions(IMF) based on HilbertHuang transform.Then,the intrinsic mode functions of high frequency is preprocessed using threshold method,and we add these IMFs with IMFs of low frequency to achieve denoising signal.Simulation experiments show that the results based on HHT are better than that of soft or hard threshold method.