Bayesian learning of the parameters of discrete and tied mixture HMMs for speech recognition.
Qiang HuoChorkin ChanChin-Hui LeePublished in: EUROSPEECH (1993)
Keyphrases
- speech recognition
- bayesian learning
- hidden markov models
- model selection
- language model
- speech synthesis
- speech recognizer
- speaker identification
- automatic speech recognition
- expectation maximization
- speech signal
- mixture model
- posterior distribution
- pattern recognition
- speech recognition systems
- maximum likelihood
- parameter estimation
- probability distribution
- hyperparameters
- speaker dependent
- em algorithm
- machine learning
- gaussian distribution
- parameter space
- closed form
- cross validation
- latent variable models
- prior knowledge
- feature selection
- data mining
- neural network