An Efficient Speech Separation Network Based on Recurrent Fusion Dilated Convolution and Channel Attention
This work addresses speech separation for practical applications, offering an incremental improvement over existing methods.
The authors tackled the problem of speech separation by proposing ARFDCN, a neural network that combines dilated convolutions, multi-scale fusion, and channel attention to address limited receptive fields and high computational costs, achieving a decent balance between performance and efficiency.
We present an efficient speech separation neural network, ARFDCN, which combines dilated convolutions, multi-scale fusion (MSF), and channel attention to overcome the limited receptive field of convolution-based networks and the high computational cost of transformer-based networks. The suggested network architecture is encoder-decoder based. By using dilated convolutions with gradually increasing dilation value to learn local and global features and fusing them at adjacent stages, the model can learn rich feature content. Meanwhile, by adding channel attention modules to the network, the model can extract channel weights, learn more important features, and thus improve its expressive power and robustness. Experimental results indicate that the model achieves a decent balance between performance and computational efficiency, making it a promising alternative to current mainstream models for practical applications.