Improved performance and generalization of minimum classification error training for continuous speech recognition.
Darryl W. PurnellElizabeth C. BothaPublished in: INTERSPEECH (2000)
Keyphrases
- minimum classification error
- continuous speech recognition
- statistical pattern recognition
- discriminative training
- maximum likelihood estimation
- classification error
- hidden markov models
- training set
- supervised learning
- data sets
- feature selection
- pairwise
- maximum likelihood
- training samples
- probabilistic interpretation