Speech Enhancement with Fullband-Subband Cross-Attention Network
Abstract
FullSubNet has shown its promising performance on speech enhancement by utilizing both fullband and subband information. However, the relationship between fullband and subband in FullSubNet is achieved by simply concatenating the output of fullband model and subband units. It only supplements the subband units with a small quantity of global information and has not considered the interaction between fullband and subband. This paper proposes a fullband-subband cross-attention (FSCA) module to interactively fuse the global and local information and applies it to FullSubNet. This new framework is called as FS-CANet. Moreover, different from FullSubNet, the proposed FS-CANet optimize the fullband extractor by temporal convolutional network (TCN) blocks to further reduce the model size. Experimental results on DNS Challenge - Interspeech 2021 dataset show that the proposed FS-CANet outperforms other state-of-the-art speech enhancement approaches, and demonstrate the effectiveness of fullband-subband cross-attention.
Keywords
Cite
@article{arxiv.2211.05432,
title = {Speech Enhancement with Fullband-Subband Cross-Attention Network},
author = {Jun Chen and Wei Rao and Zilin Wang and Zhiyong Wu and Yannan Wang and Tao Yu and Shidong Shang and Helen Meng},
journal= {arXiv preprint arXiv:2211.05432},
year = {2022}
}
Comments
Accepted by InterSpeech 2022. arXiv admin note: text overlap with arXiv:2203.12188