A modulation feature set for robust Automatic Speech Recognition in additive noise and reverberation.
Xiaoyu LiuRoozbeh SadeghianStephen A. ZahorianPublished in: ICASSP (2017)
Keyphrases
- feature set
- additive noise
- speech signal
- automatic speech recognition
- noisy environments
- speech recognition
- feature selection
- classification accuracy
- noisy images
- feature extraction
- feature vectors
- speech enhancement
- sound source
- feature space
- broadcast news
- spatial domain
- random variables
- hidden markov models
- maximum likelihood
- acoustic features
- vocal tract
- texture features
- word error rate
- speech retrieval
- speech sounds
- conversational speech
- linguistic features
- non stationary
- information retrieval
- multiplicative noise
- edge detection
- denoising
- multiscale
- machine learning