1 paper
Rui Zhou, Xian Li, Ying Fang +1
In this work, we propose Mel-FullSubNet, a single-channel Mel-spectrogram denoising and dereverberation network for improving both speech quality and automatic speech recognition (…