Research On Simultaneous Speech Detection And Magnitude Squared Spectrum Estimation Approach For Speech Enhancement

Voice communication and human-computer interaction are the most important communication methods in modern communication.However,in the process of voice communication,the voice signal is inevitably interfered by external noise,thereby hindering normal communication between people and between people and machines.The purpose of speech enhancement technology is to extract or recover pure speech signals from noisy speech as much as possible,thereby improving speech quality and improving intelligibility.Aiming at the speech enhancement problem in the background of complex noise,this paper focuses on the research and improvement of speech enhancement technology based on amplitude spectrum estimation.The main research contents are as follows:Firstly,this paper expounds the research significance and background of speech enhancement,gives the research history and development status of speech enhancement technology,and briefly introduces several commonly used speech enhancement algorithms,and gives the performance evaluation criteria of speech enhancement algorithm.Secondly,aiming at the speech enhancement problem based on the amplitude spectrum estimation theory,the paper introduces the basic principles and performance of several different short-term amplitude spectrum estimation algorithms,and briefly analyzes the advantages and disadvantages of the short-term amplitude spectrum estimation algorithm.Then,based on the short-time amplitude square spectrum estimation speech enhancement algorithm,the paper focuses on the sparse characteristics of the speech signal and the characteristics of the widely existed silent region.The speech activity detection is incorporated into the speech enhancement framework,and the fused speech is designed.The joint cost function of the squared spectrum estimation is detected,and a more efficient pure speech amplitude square spectrum estimation technique is obtained by minimizing the joint cost function.Based on this idea,the paper combines the maximum a posteriori estimation and the minimum mean square error estimation theory to propose two different speech enhancement algorithms,joint speech detection and maximum posterior amplitude square spectrum estimation algorithm and joint speech detection and amplitude square spectrum.Minimum mean square error estimation algorithm.The performance simulation and verification of the proposed two algorithms are carried out in a variety of noisy environments.The results show that the proposed algorithm can effectively improve the speech enhancement system after the combination of speech detection and amplitude square spectrum estimation.Overall performance.Finally,the paper is comprehensively summarized,and the development trend of future speech enhancement technology is prospected.