Research On Speech Enhancement Based On 2-D Microphone Array

Speech enhancement is an important branch for speech signal processing. Clean speech signals are needed in the applications such as speech recognition, speech building and speech control.It has become important to get clean signals from noisy speech. It is an effective method to solve this question by speech enhancement. This thesis researches on a 2-dimension microphone array speech enhancement method with a satisfactory result.A general speech signal model for microphone arrays has been built in the first part of this thesis. A weighted MUSIC algorithm is presented, aiming at minimizing the errors of the DOA. Based on the DOA information, a modified beamforming algorithm is applied to get a better speech enhancement result. The algorithm has been realized by MATLAB. It is shown by simulations that SNR is increased about 10dB in low SNR condition.