| 摘要 |
Disclosed herein is a noise reduction method for automatic speech recognitionl. A noise reduction method for automatic speech recognition, including: computing a magnitude spectrum (Yk(m) of a noisy speech containing a clean speech to be recognized and noise affecting the clean speech; computing a power spectrum (|Yk(m|2) of the noisy speech; computing an estimate ( |Xk (m) |2 of a power spectrum of the clean speech; computing an estimate ( Dk (m|2) of a power spectrum of the noise; computing an estimate (.xi.k(m)2) of an a priori signal-to-noise ratio as a function of the estimate (|Xk,(m)¦2) of the power spectrum of the clean speech and the estimate (|Dk(m)¦2of the power spectrum of the noise; computing an estimate (.gamma.k(m) ) of an a posteriori signal-to-noise ratio as a function of the power spectrum (|Yk(m) |2) of the noisy speech and the estimate (|Dk(m) |2) of the power spectrum of the noise; computing an attenuation gain (Gk(m)) as a function of the estimate (.xi.k(m)) of the a priori signal-to-noise ratio and the estimate (.gamma.k(m)) of the a posteriori signal-to-noise ratio; and computing an estimate (|Xk(m) ) of a magnitude spectrum of the clean speech as a function of the magnitude spectrum (|Yk(m) ) of the noisy speech and the attenuation gain (Gk(m)). Computing the estimates (.xi.k(m),.gamma.k(m)) of the a priori and the a posteriori signal-to-noise ratios includes computing a noise weighting factor (.alpha.(m)) for weighting the estimate (Dk(m)2) of the power spectrum of the noise in the computation of the estimates (.xi.k(m),.gamma.k(m)) of the a priori and the a posteriori signal-to-noise ratios; computing a spectral flooring factor (.beta.(m)) for flooring the estimates (.xi.k(m), .gamma.k(m))of the a priori and the a posteriori signal-to-noise ratios; and computing the estimates (.xi.k(m), .gamma.k(m)) of the a priori and the a posteriori signal-to-noise ratios also as a function of the noise weighting factor (.alpha.(m) and the spectral flooring factor (.beta.(m)).
|
