Improved Speech Pre-Training with Supervision-Enhanced Acoustic Unit.
Pengcheng LiGenshun WanFenglin DingHang ChenJianqing GaoJia PanCong LiuPublished in: CoRR (2022)
Keyphrases
- acoustic models
- speech recognition
- hearing impaired
- speech sounds
- acoustic features
- training set
- active learning
- emotional speech
- speech recognition systems
- training algorithm
- training process
- multi modal
- noisy environments
- speaker independent
- prosodic features
- supervised learning
- speech signal
- dialogue system
- emotion recognition
- improved algorithm
- speech synthesis
- test set
- non stationary
- endpoint detection
- hidden markov models