Research And Implementation Of Speech Enhancement Algorithm For Circular Microphone Array

With the rapid development of 5G / Internet of things technology,the requirements for speech recognition equipment in the fields of automatic speech recognition and smart home are also getting higher and higher.However,the interference of ambient and background noise can degrade the performance of such devices,so it is crucial for the study of speech enhancement algorithms.Currently,single-channel microphone speech enhancement is relatively mature,but only using the time-frequency information in the signal cannot find the direction of arrival of the signal,nor can it improve the desired signal strength.In order to overcome these shortcomings,microphone array can be used to locate the sound source of speech signal,and beamforming algorithm and echo cancellation algorithm can be used to enhance speech signal to obtain high-quality speech signal.This thesis mainly studies the speech enhancement algorithm based on circular microphone array,including sound source localization algorithm,beamforming algorithm and echo cancellation algorithm.The basics of speech enhancement algorithms for microphone arrays are studied,the properties of speech signals and the types of noise signals are analyzed,the topology of microphone arrays and their advantages and disadvantages are introduced,and two speech quality assessment strategies are described,including subjective evaluation strategies and objective evaluation strategies,the system platform adopted by OMAP-L137 in this thesis is introduced.Two classical sound source localization algorithms are studied,including the one-step sound source localization algorithm and the two-step sound source localization algorithm.Based on the Steered Response Power – Phase Transform(SRP-PHAT)algorithm,an improved sound source localization algorithm is proposed.And through the Stochastic Region Contraction(SRC)algorithm to reduce the amount of calculation.Experimental results show that the improved algorithm can accurately identify the location of sound source.Two typical beamforming algorithms are studied: fixed beamforming algorithm and adaptive beamforming algorithm.Based on Generalized Sidelobe Canceller(GSC)beamforming algorithm,an improved beamforming algorithm is proposed,which is combined with Multi Source Selection(MSS)algorithm and Dynamic Range Compression(DRC)algorithm.Experimental results show that the algorithm enhances the ability to eliminate clutter and background noise,and improves the quality of speech signal.Adaptive echo cancellation algorithms are studied,including the Least Mean Square(LMS)algorithm,the Recursive Least Square(RLS)algorithm and related algorithms.An echo cancellation algorithm based on Normalized Least Mean Square(NLMS)is proposed,and the simulation shows that the algorithm can effectively suppress the echo in the microphone signal.The performance of the hardware platform is tested.On the OMAP-L137 platform,sound source localization,beamforming and echo cancellation algorithms are used for speech signal processing and enhancement.The test results show that the signal-to-noise ratio is improved by about 46% and the signal quality is significantly improved.