A piecewise linear spectral mapping for supervised speaker adaptation.
Hiroshi MatsukotoHirowo InouePublished in: ICASSP (1992)
Keyphrases
- piecewise linear
- speaker adaptation
- speech recognition
- maximum likelihood
- automatic speech recognition
- dynamic programming
- chaotic map
- speech recognizer
- finite sets
- semi supervised
- supervised learning
- learning algorithm
- speaker independent
- principal curves
- regression algorithm
- unsupervised learning
- dimensionality reduction
- high dimensional
- image processing
- neural network
- speaker dependent