EUROSPEECH 2001 Scandinavia
7th European Conference on Speech Communication and Technology

Aalborg, Denmark
September 3-7, 2001


Sequential Noise Compensation by A Sequential Kullback Proximal Algorithm

Kaisheng Yao (1), Kuldip K. Paliwal (2), Satoshi Nakamura (1)

(1) ATR Spoken Language Translation Research Labs., Japan
(2) Griffith University, Australia

We present a sequential noise compensation method based on the sequential Kullback proximal algorithm, which uses the Kullback-Leibler divergence as a regularization function for the maximum likelihood estimation. The method is implemented as filters. In contrast to sequential noise compensation method based on the sequential EM algorithm, the convergence rate of the method and estimation error after convergence can be adjusted by a relaxation factor, where the sequential EM algorithm corresponds to the particular case of the presented algorithm. Through experiments on parameter estimation and speech recognition in noise, we verified the efficacy of the algorithm.

Full Paper

Bibliographic reference.  Yao, Kaisheng / Paliwal, Kuldip K. / Nakamura, Satoshi (2001): "Sequential noise compensation by a sequential kullback proximal algorithm", In EUROSPEECH-2001, 1139-1142.