Research On Speech Signal Processing Algorithm Based On Microphone Array

With the development of society,the number of places where people use voice to transmit information is gradually increasing.The usage of them are gradually increasing.In all kinds of speech acquisition equipment,the collected speech signals are not pure original speech signals,but noisy speech signals mixed with many interference signals or noise signals.This affects not only the voice quality,but also the overall performance of the voice processing system.Therefore,it is necessary to use a microphone array technology to recover a pure voice signal from a noisy voice signal as much as possible.The traditional microphone single-channel algorithm can remove non-coherent noise obviously,but its ability to suppress reverberation interference is very weak.While the microphone array beamforming algorithm can remove reverberation noise,but its ability is limited.Aiming at this problem,we studies the traditional microphone single-channel algorithm and microphone array algorithm,proposes a noise reduction and anti-reverberation algorithm based on the combination of improved generalized sidelobe cancellation algorithm and microphone single-channel algorithm.And we also designed a speech acquisition and processing system.The improved algorithm uses the "delay-sum" algorithm of the generalized sidelobe cancellation algorithm to obtain the voice signal of the direct array on the basis of obtaining the sound source position information,superimposes it with the signal obtained after processing by the blocking matrix and adaptive algorithm according to the weight coefficient,to generate an enhanced signal.Then,the iterative error signal obtained from the generalized sidelobe cancellation algorithm is used as the error signal to further perform the adaptive noise cancellation process for the enhanced signal,then to obtain anti-reverberation speech.Finally,the Wiener filtering method is used to suppress the residual non-coherent noise to obtain an enhanced speech signal.The experimental comparison result shows that the improved algorithm can effectively improve the speech quality under the reverberation environment.