Training data selection for improving discriminative training of acoustic models.
Shih-Hung LiuFang-Hui ChuShih-Hsiang LinHung-Shin LeeBerlin ChenPublished in: ASRU (2007)
Keyphrases
- acoustic models
- discriminative training
- hidden markov models
- speech recognition
- maximum likelihood
- probabilistic interpretation
- automatic speech recognition
- training data
- posterior probability
- speech recognizer
- support vector machine
- co training
- broadcast news
- image processing
- neural network
- training process
- high dimensional
- feature selection