Research On Correction Algorithms Of Array Manifold's Based On Microphone Array

In recent years,the field of array signals has attracted more and more scholars and experts,especially in the field of microphone arrays.The reason why the microphone array can be widely used in the speech processing system is that it can acquire the spatial position information of the sound source,thereby realizing the spatial domain filtering.The microphone array is applied to various speech enhancement and sound source localization.However,in practical applications,the effects of these algorithms are not ideal,because the actual array model of the algorithm and the ideal array model applied during simulation are used.There are big differences.The main manifestations are amplitude,phase and array position errors,which seriously affect the results of the array manifold(direction vector).Therefore,it is very important to correct the microphone array.This paper will correct the array manifold by detecting the abnormal microphone in the array and correcting the shape of the array.Firstly,the background and significance of the research are expounded,and the advantages of selecting the microphone array as the speech algorithm compared with the selection of other sensor arrays as the speech algorithm are illustrated.Then the domestic research status and the source of the topic are briefly explained.Secondly,the theoretical basis of speech signal,microphone array signal model,microphone array topology and reverberation characteristics are introduced.Then,for the problem of the error of the array pattern of the microphone array,firstly,by studying the traditional microphone correction algorithm(active correction and self-correction algorithm),it can be known that the traditional active correction algorithm and self-correction algorithm are hypothetical microphone arrays.There are no abnormalities in the array elements.In view of the possibility of abnormal microphones,this paper studies the method of detecting abnormal microphones and detects abnormal microphones.This method is used as the correction of the microphone array manifold.Then,for the case where the array model may cause errors,the actual shape of the microphone array is corrected by studying the shape correction method of the microphone array,thereby achieving the purpose of correcting the array manifold.Finally,the corrected array manifold is applied to the speech enhancement algorithm,and the speech signals collected under the presence or absence of reverberation conditions are compared as sound inputs,and the file transmission is realized by the TCP/IP protocol,and the collected speech is obtained.The signal is saved on the computer through the Labview program,and then sent by the computer as a client to the DSP server.After the DSP processes the algorithm,the DSP is used to build a voice enhanced prototype system that can run in the real environment.The algorithm is applied to the algorithm.In the actual system,the results of the final test prove that the enhancement effect of speech is obvious,indicating that the speech enhancement algorithm works well on the platform.