Far-field Microphone Array Speech Enhancement And System Implementation

With development of social economy and science-technology,requirement of security products is extending from the traditional video images surveillance to video-voice collection.In recent years,digital signal processing technology and array signal processing technology have been developed rapidly to improve the performance of far-field voice collection.There is a tendency to gradually replace single channel speech signal pickup in field of intelligent home and human-machine interaction.In this paper,a speech enhancement system is designed under security background to pick up speech signal in far-field.The main work of this thesis is as follows:(1)Summarizing the traditional sound source localization technology of TDOA,MUSIC,SRP-PHAT and DS,and comparing the advantages and disadvantages of each algorithm.A far-field compressed sensing microphone array DOA estimation approach is proposed and evaluated with experimental data.(2)Introducing the basic principle of fixed beamforming speech enhancement,GSC and Zelinski post-filter speech enhancement,in view of the limitation of hardware implementation the fixed beamforming algorithm is adopted as the system enhancement algorithm.(3)Based on STM32 embedded digital signal processor,a far-field microphone array speech enhancement system is designed for far-field acoustics surveillance.The far-field source speech signal picked up experiment is carried out to verify the feasibility of the system.