Incorporating the human hearing properties in the GSVD-based approach for speech enhancement

A perceptually motivated subspace-based approach is proposed for the enhancement of speech corrupted by colored noise. The proposed approach takes into account the frequency masking properties of the human auditory system and reduces the perceptual effect of the residual noise. A new GSVD-based approach, called perceptually constrained GSVD (PCGSVD), is proposed, where the difference from the previously proposed truncated QSVD-based approach is clearly indicated. Closed-form solution is obtained for the PCGSVD enhancement approach. Very carefully performed objective evaluations and subjective tests show that the PCGSVD approach proposed here can offer improved speech quality, and intelligibility in the case of stationary and nonstationary noises, especially when the additive noise is nonwhite.