De-noising Study Of Speech Signal Based On Wavelet Method

Speech signal de-noising is an important component in the signal processing. Wavelet can analyze signal at time domain and frequency simultaneously, so it can de-noise effectively. Because the mother wavelets is a gather, which mother wavelet can reach best de-noising effect at practical application will to be resolve. Then, the election of thresholds has immediate relation to the result of de-noising. Some wavelet coefficients can't be set zero when the threshold is undersize and parts of noises are retained. Some useful signals will be reduced if the threshold is up-size. These cases may degrade the de-noising effect.Two questions are mainly studied in this thesis. In this article five layers of db3 wavelet is selected as mother wavelet, four threshold selection rules and different rescaling methods are used to analysis the noisy speech signal. A novel threshold denoising method named adaptive wavelet coefficients subtraction is proposed based on short-time Fourier transformation. The proposed method is based on SURE unbias estimation and the most optimal threshhold parameters can be researched adaptively satisfy the requirement of time variable signal in real-time processing. At last, the proposed method is simulated using Matlab7.0 software, and the results are compared with that of spectrum subtractive method. The results of simulation prove that adaptive wavelet coefficients subtraction method can effectively restrain noises and improve SNR dramaticly.