Using Separate Losses for Speech and Noise in Mask-Based Speech Enhancement.
Ziyi XuSamy ElshamyTim FingscheidtPublished in: ICASSP (2020)
Keyphrases
- speech enhancement
- noisy environments
- noise reduction
- signal to noise ratio
- speech signal
- single channel
- noisy speech
- linear prediction
- vocal tract
- speech recognition
- spectral subtraction
- background noise
- additive noise
- wiener filter
- smoothing algorithm
- multi channel
- edge preserving
- speaker identification
- edge detection
- information retrieval
- non stationary
- multiresolution