Adaptive Speech Enhancement Algorithm Based on First-order Differential Microphone Array

To solve the problem of directional interference noise in portable voice communication equipment, a dual-channel adaptive speech enhancement algorithm combined with the first-order differential microphone array and variable-step-size frequency-domain Least Mean Square (LMS) algorithm is proposed. First, the speech channel signal and noise channel signal were obtained by using the first-order differential array process the noise signal by microphone. Then, the residual noise in the speech channel signal was further eliminate by using the two channel signals combined with the variable-step-size frequency-domain LMS to estimate the optimal step length to control the update of the frequency-domain weight coefficient of the adaptive filter, thereby improve speech quality and intelligibility. The simulation results show that the algorithm overcomes the influence of directional noise effectively, and the objective evaluation of speech quality and the effect of noise reduction are better than the comparison algorithm.