Speech Enhancement Algorithm based on Signal Subspace with Low Signal-to-Noise Ratio

Compared with other classic speech enhancement algorithm, buit-in pre-whitening subspacealgorithm has a good denoising effect, but in a low signal-to-noise ratio environment, there is still a large amount of residual noise. For this, the method that combining signal subspace optimization andwiener filtering is put forward to improve the speech quality. Simulation results show that underseveral common background noises with low signal-to-noise ratio, the performance of the proposedapproach ismore excellent, it can effectively restrain background noise and improve the speech quality, meanwhile, in some extent, guarantee the speech intelligibility, compared with the traditional spectral-subtraction, wiener filtering method, buit-in pre-whitening subspacealgorithm.