A non-linear filtering approach to stochastic training of the articulatory-acoustic mapping using the EM algorithm.
Gordon RamsayPublished in: ICSLP (1996)
Keyphrases
- em algorithm
- expectation maximization
- mixture model
- maximum likelihood
- maximum likelihood estimation
- gaussian mixture model
- generative model
- parameter estimation
- likelihood function
- expectation maximisation
- hidden variables
- training set
- incomplete data
- gibbs sampling
- log likelihood
- acoustic features
- finite mixture model
- probabilistic principal component analysis
- matrix factorisation
- gaussian mixture
- model based clustering
- finite mixture models
- maximum a posteriori
- mixture distribution
- speech recognition
- unsupervised learning
- likelihood maximization
- model selection
- feature space