A maximum a posteriori approach to speaker adaptation using the trended hidden Markov model.
Rathinavelu ChengalvarayanLi DengPublished in: IEEE Trans. Speech Audio Process. (2001)
Keyphrases
- hidden markov models
- maximum a posteriori
- speaker adaptation
- speech recognition
- maximum likelihood
- speaker independent
- automatic speech recognition
- map estimation
- markov random field
- image reconstruction
- bayesian framework
- speaker dependent
- em algorithm
- expectation maximization
- hyperparameters
- speech recognizer
- posterior distribution
- hidden states
- energy function
- noisy environments
- speech synthesis
- markov networks
- feature selection
- speaker identification
- conditional random fields
- speech recognition systems
- speech signal
- broadcast news
- mixture model
- bayesian networks