Fast speaker adaptation using triple diagonal and shared block diagonal transform matrices.
Guo-Hong DingBo XuJuha Iso-SipiläYang CaoPublished in: ICASSP (1) (2003)
Keyphrases
- block diagonal
- speaker adaptation
- semidefinite programming
- speech recognition
- linear programming problems
- kernel matrix
- augmented lagrangian
- maximum likelihood
- automatic speech recognition
- covariance matrix
- hidden markov models
- affinity matrix
- machine learning
- spectral clustering
- linear program
- linear programming
- pattern recognition