Research on generalized sidelobe canceller based on modified wavelet threshold function

In the microphone array speech enhancement, The coherent noise has a great influence on the accuracy and speech leakage of the delay estimation in the beamforming algorithm. In this paper, the wavelet threshold denoising method is effectively combined with the beamforming algorithm to acting on every signal of microphone array. It can eliminate coherent noise in the signal to reduce noise-induced delay errors and leakage. In the wavelet threshold denoising algorithm, the threshold function directly determines the enhancement effect. So in this paper, we propose an improved continuous threshold function with adjustment parameters and the wavelet coefficients of speech signal with noisy could be processed in the wavelet domain. Speech signal after being eliminate can be obtained through reconstructing. The adaptive beamformer of the generalized sidelobe canceller uses the improved LMS algorithm to speed up the updating speed of the weight coefficient. The simulation results show that compared with the traditional generalized sidelobe canceller, the proposed algorithm improves the signal-to-noise ratio and intelligibility of the speech signal.