Stream Attention Based U-Net for L3DAS23 Challenge

Honglong Wang; Yanjie Fu; Junjie Li; Meng Ge; Longbiao Wang; Xinyuan Qian

ICASSP 2023accepted0 citations

Stream Attention Based U-Net for L3DAS23 Challenge

Honglong Wang, Yanjie Fu, Junjie Li, Meng Ge, Longbiao Wang, Xinyuan Qian

Abstract

Machine learning applications of 3D audio are gaining increasing interest in recent years. In this paper, we propose a stream attention based U-Net to remove background noise and reverberation based on ICASSP Signal Processing Grand Challenge 2023: L3DAS23 Challenge<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> Audio-only track task1 3D Speech Enhancement. Results show that proposed method achieves superior performance than the official baseline model.

BibTeX

@inproceedings{icassp2023_streamattentionb,
  title = {Stream Attention Based U-Net for L3DAS23 Challenge},
  author = {Honglong Wang and Yanjie Fu and Junjie Li and Meng Ge and Longbiao Wang and Xinyuan Qian},
  booktitle = {ICASSP 2023},
  year = {2023}
}