Maximum a posteriori eigenvoice speaker adaptation for Korean connected digit recognition.
Hyung Bae JeonDong Kook KimPublished in: INTERSPEECH (2004)
Keyphrases
- maximum a posteriori
- digit recognition
- speaker adaptation
- speaker independent
- maximum likelihood
- map estimation
- hand written
- markov random field
- image reconstruction
- noisy environments
- speech recognition
- em algorithm
- audio visual
- energy function
- support vector machine
- automatic speech recognition
- expectation maximization
- mixture model
- image segmentation
- generative model
- graph cuts
- speech recognizer
- image classification