On Loss Functions for Supervised Monaural Time-Domain Speech Enhancement.
Morten KolbækZheng-Hua TanSøren Holdt JensenJesper JensenPublished in: CoRR (2019)
Keyphrases
- loss function
- speech enhancement
- noise reduction
- noisy environments
- single channel
- signal to noise ratio
- frequency domain
- pairwise
- semi supervised
- learning algorithm
- support vector
- speech signal
- loss minimization
- linear prediction
- multi channel
- supervised learning
- risk minimization
- learning to rank
- vocal tract
- convex loss functions
- wiener filter
- additive noise
- reproducing kernel hilbert space
- speech recognition
- edge detection
- smoothing algorithm
- information retrieval
- feature extraction
- feature selection
- spatial domain
- lossless compression
- image restoration
- multi class
- machine learning