Pinched lattice minimum Bayes risk discriminative training for large vocabulary continuous speech recognition.
Vlasios DoumpiotisWilliam ByrnePublished in: INTERSPEECH (2004)
Keyphrases
- bayes risk
- discriminative training
- posterior probability
- automatic speech recognition
- hidden markov models
- upper and lower bounds
- loss function
- probabilistic interpretation
- maximum likelihood
- speech recognition
- bayesian networks
- bayesian framework
- generative model
- reproducing kernel hilbert space
- probability distribution
- support vector machine
- probabilistic model
- distortion measure
- reinforcement learning
- probability density function
- mixture model
- regularization parameter
- vector quantizer