Speech enhancement based on Laplacian model and masking
XU Cui-xiang, MA Jian-fen · Computer Engineering and Applications Journal · 2010
An effective approach for attenuating acoustic noise and mitigating speech distortion is proposed.First,MMSE method is analysed when the clean speech is modeled by a Laplacian distribution and the noise is modeled by a Gaussian distribution.Then,human perceptual auditory masking threshold is incorporated into this approach when the threshold of spectral amplitude of enhanced speech is computed.The experiment result evaluated by objective measure shows the proposed method can achieve a more significant noise reduction and reduce the chances of speech distortion.