International Symposium on Chinese Spoken Language Processing (ISCSLP 2002)

Taipei, Taiwan
August 23-24, 2002

Integration of Model Adaptation and Missing Feature Theory for Robust Speech Recognition

An-Tze Yu, Hsiao-Chuan Wang

Department of Electrical Engineering, National Tsing Hua University, Hsinchu, Taiwan

This paper presents a method to integrate the model adaptation technique and the missing feature theory for robust speech recognition. The recognition performance can be further improved as comparing with that utilizing only a single approach. The basic idea comes from that the integration of two techniques will compensate two improper assumptions made separately by these two techniques. The evaluation of the proposed method on the task of isolated Mandarin digits recognition shows that the system robustness is upgraded substantially.


Full Paper

Bibliographic reference.  Yu, An-Tze / Wang, Hsiao-Chuan (2002): "Integration of model adaptation and missing feature theory for robust speech recognition", In ISCSLP 2002, paper 54.