Perceptual Contrast Stretching on Target Feature for Speech Enhancement
This work addresses speech enhancement for audio processing applications, presenting an incremental improvement by integrating perceptual weighting into training.
The paper tackles speech enhancement by proposing a perceptual contrast stretching approach to modify target features, achieving state-of-the-art performance with PESQ scores of 3.07 for causal and 3.35 for noncausal tasks on the VoiceBank-DEMAND dataset.
Speech enhancement (SE) performance has improved considerably owing to the use of deep learning models as a base function. Herein, we propose a perceptual contrast stretching (PCS) approach to further improve SE performance. The PCS is derived based on the critical band importance function and is applied to modify the targets of the SE model. Specifically, the contrast of target features is stretched based on perceptual importance, thereby improving the overall SE performance. Compared with post-processing-based implementations, incorporating PCS into the training phase preserves performance and reduces online computation. Notably, PCS can be combined with different SE model architectures and training criteria. Furthermore, PCS does not affect the causality or convergence of SE model training. Experimental results on the VoiceBank-DEMAND dataset show that the proposed method can achieve state-of-the-art performance on both causal (PESQ score = 3.07) and noncausal (PESQ score = 3.35) SE tasks.