A variational EM algorithm for learning eigenvoice parameters in mixed signals.

Ron J. Weiss Daniel P. W. Ellis

Published in: ICASSP (2009)

Keyphrases

em algorithm
expectation maximization
maximum likelihood
parameter estimation
maximum likelihood estimation
parameter learning
hidden variables

gaussian mixture
gaussian mixture model
mixture model
likelihood function
likelihood maximization
generative model

learning algorithm
expectation maximisation
unsupervised learning
reinforcement learning
bayesian networks
log likelihood function

closed form
probability density function
hyperparameters
probabilistic model
prior knowledge
image segmentation