Discriminative linear transforms for feature normalization and speaker adaptation in HMM estimation.
Stavros TsakalidisVlasios DoumpiotisWilliam ByrnePublished in: INTERSPEECH (2002)
Keyphrases
- speaker adaptation
- speech recognition
- hidden markov models
- maximum likelihood
- speaker independent
- automatic speech recognition
- discriminative features
- semi supervised
- speech recognizer
- speaker dependent
- image features
- feature selection
- feature set
- classification accuracy
- speaker identification
- feature extraction
- machine learning
- neural network