Cross channel interaction based ECA-Net using gated recurrent convolutional network for speech enhancement

Recently channel attention mechanism playing a major role in improving the performance of deep convolution neural networks. Even though there is an improvement in the performance, but there is an increase in complexity of the model network. It is difficult for CNN alone to correctly model the long-range dependencies of speech signals. The local receptive field of the convolution limits the model’s ability to capture long-range dependencies across input sequences. Moreover, LSTMs are easily prone to the problem of overfitting, and it also requires a long time to train. To overcome the above drawbacks in the proposed model, dilated dense blocks and GRUs are introduced. Based on multi-task learning, we propose a gated convolutional recurrent network with efficient channel attention (GCRN-ECA) for complex spectral mapping, which amounts to a causal system for monaural speech enhancement. Each layer in encoder and decoder consists of dense block. Complex spectral mapping seeks to predict the real and imaginary spectrograms of clear speech based on those of noisy speech, thereby enhancing both the magnitude and phase responses of the speech. The advantage of dilated convolutions in dense block is the receptive field increases with increasing dilation rates, which are used to capture long-range speech contexts. And the dense connectivity provides a feature map with more precise target information by passing through multiple layers. To represent the correlation between neighboring noisy speech frames, a two Layer GRU is added in the bottleneck, which has the advantage of increased training speed because of its simpler architecture. GRU captures the long-range dependencies across input sequences. The advantage of GRU is that it is easier to modify and doesn’t require memory units, which means it can train faster than LSTM. The ECA module can implement cross-channel interaction without dimensionality reduction. An appropriate cross-channel interaction can preserve performance while significantly decreasing model complexity. Our results reveal that the proposed GCRN-ECA outperforms existing baselines in terms of quality and intelligibility. The proposed model increases the average PESQ and STOI scores by 35.9% and 18.1%, respectively, for the Common voice dataset, and by 35.2% and 4%, respectively, for the VCTK dataset compared to noisy speech.