Research On Fast Denoising Of Speech Signal Based On Adaptive Wavelet In Turbulent Channel

In the turbulent channel,noise reduction technology is one of the key technologies of the transmission system when it uses a laser to transmit voice signals.This thesis relies on the research project of wireless laser detection in the technology research projects,and implements the adaptive wavelet fast denoising on the noisy speech signal collected on the laser listening platform.The denoising algorithm is verified and the corresponding algorithm is given.The thesis adopts theoretical analysis,concrete realization and test verification to study the adaptive wavelet denoising technology of speech signal.In the theoretical analysis stage,first of all,through analyzing the research status of speech denoising technology both at domestic and overseas,combining the characteristics of speech signals and noise signals in turbulence channels,the speech denoising algorithm is studied,and the performance evaluation criteria of speech denoising algorithms are formed.Then the principle and design of the conventional speech noise reduction method are analyzed,Wavelet transform is selected as the noise reduction algorithm of this paper.Based on the original wavelet transform noise reduction algorithm,an adaptive wavelet fast noise reduction algorithm is proposed.In the specific implementation stage,the common audio file format is analyzed first,and it is determined that the audio file is saved in the WAV file format.Then by comparing the advantages and disadvantages of the traditional wavelet denoising algorithm and the adaptive wavelet fast denoising algorithm,the realization steps of this thesis 's algorithm are given,and the determination methods of wavelet basis,wavelet decomposition layer number,threshold function and threshold value are analyzed.Select the most suitable parameter for speech denoising.Finally,by analyzing the CUDA programming language concept on the GPU,the CPU and GPU are used to implement the algorithm.In the test verification stage,the speech denoising performance and time performance of the adaptive wavelet fast denoising algorithm were tested using the voice signal data collected by the laboratory wireless laser listening platform.The results show that the signal-to-noise ratio(SNR)improves by 6dB;the length of speech denoising is greatly shortened,indicating that the adaptive wavelet fast denoising.algorithm used in this paper can effectively improve the speech quality of noisy speech signals in turbulent channels.