Time-Frequency Attention for Monaural Speech Enhancement
This work addresses the problem of accurate speech enhancement for audio processing applications, representing an incremental advance by integrating attention into existing network architectures.
The paper tackles speech enhancement by introducing a time-frequency attention module that provides differentiated weights to spectral components, resulting in significant improvements across five objective metrics with minimal parameter overhead.
Most studies on speech enhancement generally don't consider the energy distribution of speech in time-frequency (T-F) representation, which is important for accurate prediction of mask or spectra. In this paper, we present a simple yet effective T-F attention (TFA) module, where a 2-D attention map is produced to provide differentiated weights to the spectral components of T-F representation. To validate the effectiveness of our proposed TFA module, we use the residual temporal convolution network (ResTCN) as the backbone network and conduct extensive experiments on two commonly used training targets. Our experiments demonstrate that applying our TFA module significantly improves the performance in terms of five objective evaluation metrics with negligible parameter overhead. The evaluation results show that the proposed ResTCN with the TFA module (ResTCN+TFA) consistently outperforms other baselines by a large margin.