A hybrid speech recognition system using HMMs with an LVQ-trained codebook.
Hitoshi IwamidaShigeru KatagiriErik McDermottYoh'ichi TohkuraPublished in: ICASSP (1990)
Keyphrases
- speech recognition
- hidden markov models
- multi stream
- training set
- vector quantization
- speech signal
- discriminative training
- phoneme recognition
- image representation
- learning vector quantization
- supervised learning
- automatic speech recognition
- speech synthesis
- feature vectors
- expectation maximization
- speech recognizer
- spoken language
- speaker identification
- endpoint detection
- maximum likelihood
- broadcast news
- dialogue system
- vector quantizer
- decision trees
- audio visual
- svm classifier
- image features