MAP Estimation of Whole-Word Acoustic Models with Dictionary Priors.
Keith KintzleyAren JansenHynek HermanskyPublished in: INTERSPEECH (2012)
Keyphrases
- map estimation
- maximum a posteriori
- image super resolution
- acoustic models
- speech recognizer
- maximum likelihood
- speech recognition
- markov random field
- noisy images
- em algorithm
- image reconstruction
- parallel corpus
- spoken language
- maximum a posterior
- sparse representation
- energy function
- bayesian framework
- linguistic knowledge
- speech recognition systems
- hidden markov models
- expectation maximization
- n gram
- discriminative training
- image sequences
- graphical models
- machine learning