English

DESNet: A Multi-channel Network for Simultaneous Speech Dereverberation, Enhancement and Separation

Sound 2020-11-17 v3 Audio and Speech Processing

Abstract

In this paper, we propose a multi-channel network for simultaneous speech dereverberation, enhancement and separation (DESNet). To enable gradient propagation and joint optimization, we adopt the attentional selection mechanism of the multi-channel features, which is originally proposed in end-to-end unmixing, fixed-beamforming and extraction (E2E-UFE) structure. Furthermore, the novel deep complex convolutional recurrent network (DCCRN) is used as the structure of the speech unmixing and the neural network based weighted prediction error (WPE) is cascaded beforehand for speech dereverberation. We also introduce the staged SNR strategy and symphonic loss for the training of the network to further improve the final performance. Experiments show that in non-dereverberated case, the proposed DESNet outperforms DCCRN and most state-of-the-art structures in speech enhancement and separation, while in dereverberated scenario, DESNet also shows improvements over the cascaded WPE-DCCRN networks.

Keywords

Cite

@article{arxiv.2011.02131,
  title  = {DESNet: A Multi-channel Network for Simultaneous Speech Dereverberation, Enhancement and Separation},
  author = {Yihui Fu and Jian Wu and Yanxin Hu and Mengtao Xing and Lei Xie},
  journal= {arXiv preprint arXiv:2011.02131},
  year   = {2020}
}

Comments

Accepted at IEEE SLT 2021

R2 v1 2026-06-23T19:54:20.172Z