An improved method for speech signal processing based on adaptive filter

Based on the adaptive filtering mode, we discussed and analyzed how to improve the quality of microphone array speech enhancement and reduce the complexity of the algorithm. This paper described a new variable step least mean square (LMS) algorithm which could reduce the effect of input noise to step factor, improve the performance of the algorithm. The algorithm smoothed the step factor in time domain when they were in the process of iteration so that we could eliminate the influence on the overall trend when updating the step. Also we used the Sigmoid function to control direction of the update trend. Our results validated the new variable step algorithm has good performance.