Improving Accented Speech Recognition with Multi-Domain Training.
Lucas MaisonYannick EstèvePublished in: ICASSP (2023)
Keyphrases
- speech recognition
- multi domain
- wall street journal corpus
- isolated word
- acoustic models
- hidden markov models
- cross domain
- language model
- speech synthesis
- automatic speech recognition
- speaker identification
- domain specific
- speech recognition systems
- speech processing
- noisy environments
- speech signal
- training set
- speech recognizer
- speaker independent
- speech recognition technology
- speaker dependent
- data mining
- heterogeneous networks
- supervised learning
- bayesian networks
- discriminative training
- learning styles
- feature space