431380

Kernel Eigenspace-based MLLR Adaptation Using Multiple Regression Classes

Hsiao, Roger ; Mak, Brian

Dept. of Comput. Sci., Hong Kong Univ. of Sci. & Technol., China

1

2005

March 18-23, 2005

985

988

adaptive signal processing; eigenvalues and eigenfunctions; maximum likelihood estimation; optimisation; principal component analysis; regression analysis; speech recognition; KEMLLR; KEV; KPCA; eigenmatrices; eigenvoice-based adaptation methods; kernel eigenspace-based MLLR adaptation; kernel eigenvoice adaptation; kernel principal component analysis; kernel-induced high dimensional feature space; multiple regression classes; quasi-Newton BFGS optimization algorithm; speaker-dependent HMM; speaker-dependent MLLR transformation matrices; Application software; Bayesian methods; Computer science; Hidden Markov models; Kernel; Loudspeakers; Maximum likelihood linear regression; Principal component analysis; Space technology; Speech;

English

ieee

Acoustics, Speech, and Signal Processing, 2005. Proceedings. (ICASSP '05). IEEE International Conference on

1520-6149

0-7803-8874-7

conf

10.1109/ICASSP.2005.1415281

1415281