Speaker time-drifting adaptation using trajectory mixture hidden Markov models.
Jian SuHaizhou LiJean-Paul HatonKai-Tat NgPublished in: ICASSP (1996)
Keyphrases
- hidden markov models
- speech recognition
- speaker adaptation
- automatic speech recognition
- speaker independent
- mixture distributions
- latent variable models
- markov models
- mixture model
- maximum likelihood
- sequential data
- visual speech
- phoneme recognition
- hidden state
- gesture recognition
- speaker verification
- speaker dependent
- handwritten text recognition
- acoustic models
- continuous hidden markov models
- automatic speech recognition systems
- sequence classification
- speaker identification
- handwriting recognition
- noisy environments
- markov model
- gaussian mixture model
- conditional random fields
- expectation maximization
- keyword spotting
- broadcast news
- visual speech recognition
- estimate the model parameters
- spatio temporal