Vocal tract length normalization using rapid maximum-likelihood estimation for speech recognition.
Tadashi EmoriKoichi ShinodaPublished in: Systems and Computers in Japan (2002)
Keyphrases
- vocal tract
- maximum likelihood estimation
- speech recognition
- speech signal
- speech synthesis
- length normalization
- speaker adaptation
- maximum likelihood
- em algorithm
- automatic speech recognition
- hidden markov models
- parameter estimation
- speech enhancement
- probability distribution
- language model
- expectation maximization
- spectral analysis
- pattern recognition
- linear prediction
- feature vectors
- speaker independent
- mixture model
- text to speech
- multimedia
- information retrieval