TY - JOUR
T1 - Online speaker adaptation based on quasi-Bayes linear regression
AU - Chien, Jen-Tzung
AU - Huang, C. H.
PY - 2001
Y1 - 2001
N2 - This paper presents an online/sequential linear regression adaptation framework for hidden Markov model (HMM) based speech recognition. Our attempt is to sequentially improve speaker-independent (SI) speech recognizer to meet nonstationary environments via linear regression adaptation of SI HMM's. A quasi-Bayes linear regression (QBLR) algorithm is developed to execute online adaptation where the regression matrix is estimated using QB theory. In the estimation, we moderately specify the prior density of regression matrix as a matrix variate normal distribution and exactly derive the pooled posterior density belonging to the same distribution family. Accordingly, the optimal regression matrix can be easily calculated. Also, the reproducible prior/posterior density pair provides meaningful mechanism for sequential learning of prior statistics. At each sequential epoch, only the updated prior statistics and the current observed data are required for adaptation. In general, the proposed QBLR is universal and can be reduced to well-known maximum likelihood linear regression (MLLR) and maximum a posteriori linear regression (MAPLR). Experiments show that the QBLR is effective for speaker adaptation in car environments.
AB - This paper presents an online/sequential linear regression adaptation framework for hidden Markov model (HMM) based speech recognition. Our attempt is to sequentially improve speaker-independent (SI) speech recognizer to meet nonstationary environments via linear regression adaptation of SI HMM's. A quasi-Bayes linear regression (QBLR) algorithm is developed to execute online adaptation where the regression matrix is estimated using QB theory. In the estimation, we moderately specify the prior density of regression matrix as a matrix variate normal distribution and exactly derive the pooled posterior density belonging to the same distribution family. Accordingly, the optimal regression matrix can be easily calculated. Also, the reproducible prior/posterior density pair provides meaningful mechanism for sequential learning of prior statistics. At each sequential epoch, only the updated prior statistics and the current observed data are required for adaptation. In general, the proposed QBLR is universal and can be reduced to well-known maximum likelihood linear regression (MLLR) and maximum a posteriori linear regression (MAPLR). Experiments show that the QBLR is effective for speaker adaptation in car environments.
UR - http://www.scopus.com/inward/record.url?scp=0034841235&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2001.940834
DO - 10.1109/ICASSP.2001.940834
M3 - Conference article
AN - SCOPUS:0034841235
SN - 1520-6149
VL - 1
SP - 329
EP - 332
JO - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
JF - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
T2 - 2001 IEEE Interntional Conference on Acoustics, Speech, and Signal Processing
Y2 - 7 May 2001 through 11 May 2001
ER -