Techniques in rapid unsupervised speaker adaptation based on HMM-Sufficient Statistics.
Randy GomezTomoki TodaHiroshi SaruwatariKiyohiro ShikanoPublished in: Speech Commun. (2009)
Keyphrases
- speaker adaptation
- sufficient statistics
- speech recognition
- maximum likelihood
- automatic speech recognition
- hidden markov models
- speaker independent
- speech recognizer
- k means
- data points
- speaker dependent
- unsupervised learning
- transition probabilities
- pattern recognition
- expectation maximization
- semi supervised
- language model
- hidden variables
- gaussian mixture model
- supervised learning
- image processing
- machine learning
- speech signal
- statistical queries
- speech synthesis
- vocal tract
- training data