Rapid speaker adaptation with speaker adaptive training and non-negative matrix factorization.
Xueru ZhangKris DemuynckHugo Van hammePublished in: ICASSP (2011)
Keyphrases
- speaker adaptation
- negative matrix factorization
- speech recognition
- maximum likelihood
- automatic speech recognition
- matrix factorization
- speaker independent
- speaker dependent
- speech recognizer
- document clustering
- principal component analysis
- hidden markov models
- sparse representation
- spectral clustering
- language model
- training set
- non stationary
- missing data
- em algorithm
- speech signal
- speech synthesis
- pattern recognition
- multiscale
- information retrieval