MetricGAN-U: Unsupervised Speech Enhancement/ Dereverberation Based Only on Noisy/ Reverberated Speech.
Szu-Wei FuCheng YuKuo-Hsuan HungMirco RavanelliYu TsaoPublished in: ICASSP (2022)
Keyphrases
- speech enhancement
- noisy environments
- noise reduction
- speech signal
- speech recognition
- noisy speech
- speaker identification
- background noise
- signal to noise ratio
- linear prediction
- automatic speech recognition
- vocal tract
- single channel
- wiener filter
- image sequences
- multi channel
- non stationary
- optical flow
- smoothing algorithm
- pattern recognition
- feature extraction