Abstract:

Be aimed at the problem that reference output signal is required in adaptive speech enhancement, the paper proposed two new algorithms: It just uses an input value of observation voice noise, it avoids using a delay quantity of input signal. This algorithm detected an expected speech signal of derived from spectral subtraction, this expected signal is used as reference signal of the output filter; It choose differences as the error signal between the referenced signal and the output filter, The filters of reconstruction are revised directly by filter corrective value based on LMS and RLS of weigh. This algorithm adopts the way with block technology and symbol technology. It reduces the computation of corrective weight and improves the convergence speed. The simulation experiment shows that the algorithm can effectively enhance speech in all kinds of SNR, and it is superior to Spectral Subtraction. The result shows that the enhancement method is satisfaction in the 3 dB.

Key words: adaptive speech enhancement;spectral subtraction;LMS algorithm;RLS algorithm;filter weight coefficient