Iterative training of a DPGMM-HMM acoustic unit recognizer in a zero resource scenario.
Michael HeckSakriani SaktiSatoshi NakamuraPublished in: SLT (2016)
Keyphrases
- hidden markov models
- speech recognition
- acoustic models
- speaker independent
- discriminative training
- baum welch
- training set
- iterative learning
- training stage
- speech recognizer
- multi stream
- training process
- resource allocation
- speech recognition systems
- resource constraints
- speaker identification
- iterative methods
- test set
- decision trees
- speech recognizers
- resource management
- gesture recognition
- information resources
- training samples
- pattern recognition
- face recognition