A SVD-Based Beamforming Speech Enhancement Algorithm

摘要

In this paper,a singular value decomposition (SVD) weighting beamforming algorithm(SWBF) is proposed for enhancing multimicrophone speech signals degraded by additive noise. First the algorithm adopts the crosspower spectrum phase(CSP) algorithm to compensate the time delay of the signal received by the microphone array. Second, the algorithm adopts Jacobi algorithm to carry out SVD for the input matrix, and then use the singular value as the weighting value to get the weighting sum, which is the output enhancement speech. The simulation shows, in the addictive uncorrelative noise, the output signal to noise rat io(SNR) has obvious improvement compared with fixed weighting beamforming(FWBF) speech enhancement algorithm.