Research The Speech Denoising Method Based On Wiener Filtering And Wavelet Threshold

Speech denoising which is extracting the important speech signals we need from the noisy noise by technical means,and to minimize the impact of the noise and try our best to separate the pure original speech signal.Therefore,the significance of speech denoising is to reduce the noise signal in noisy speech as much as possible and make the speech more pure and easy to receive.First,the paper first introduce the research significance of speech denoising,the history and development of speech denoising,and gives the research contents and structure of this paper,and introduces the characteristics of human ear perception,noise classification,speech signal features and information content.Then,introduce the signal model of speech denoising and the commonly used denoising methods are introduced,and the SNR is introduced as the evaluation standard of speech denoising.Secondly,this paper introduces the basic principle of Wiener filter and the concrete steps of the realization of Wiener filter algorithm,and carries on the simulation experiment to the Wiener filter algorithm.Wiener filter is applied to the pure speech signal with different signal-to-noise ratio and different types of noise signal,and obtain the waveform of the filtered speech signal as well as calculate the signal-to-noise ratio of the noisy speech signal after Wiener filtering.The filtering effect of Wiener filtering algorithm for different types of noise superimposed with different signal-to-noise ratio is analyzed.Thirdly,this paper focuses on the basic theory of wavelet transform,several common wavelet functions and the main considerations of selecting wavelet functions,analyzes in detail the principle of wavelet threshold denoising,the selection of threshold functions and four threshold rules.Through simulation experiments,compare the denoising effects of four threshold rules under different decomposition layers.Finally,this paper proposes a Wiener-wavelet threshold denoising method,which is also the innovation of this paper.The method can filter most of the noise by using the minimum mean square error of wiener filter,and then carry out the wavelet threshold quadratic denoising.This method can effectively avoid using wavelet threshold to filter some useful speech signals,which reduces the speech quality after denoising.Through four simulation experiments,wavelet threshold method and Wiener wavelet threshold method are used to denoise the noisy speech signals with different signal-to-noise ratio and different noise types,and compares their denoising effects.The experimental results show that the signal-to-noise ratio(SNR)of speech signal processed by Wiener wavelet threshold method is greatly improved by wavelet threshold method,which indicates that Wiener wavelet threshold method has better denoising effect than wavelet threshold method.The denoising of noisy speech signal is more effective.