First International Conference on Spoken Language Processing (ICSLP 90)

Kobe, Japan
November 18-22, 1990

A Comparative Study of Speaker Adaptation Methods for HMM-Based Speech Recognition

M. W. Koo, Chong Kwan Un, Hwang Soo Lee, J. M. Koo, H. R. Kim

Communications Research Laboratory, Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Seoul, Korea

In this paper, we compare the performances of speaker adaptation algorithms which consist of two stages of processing for an HMM-based speech recognition system. We compare three kinds of VQ adaptation methods which may be used in the first stage to reduce the total distortion error for a new speaker; label prototype adaptation, adaptation with a codebook from adaptation speech itself, and adaptation with a mapped codebook. We then compare the performance of four kinds of HMM parameter adaptation methods which may be used in the second stage to transform HMM parameters for a new speaker; adaptation by the Viterbi algorithm, that by the DTW algorithm, that by the iterative alignment algorithm, and that by the fuzzy mapped codebook algorithm. The results show that adaptation based on the mapped codebook and the DTW algorithm yields the highest accuracy in an HMM-based speech recognition system.

Full Paper

Bibliographic reference.  Koo, M. W. / Un, Chong Kwan / Lee, Hwang Soo / Koo, J. M. / Kim, H. R. (1990): "A comparative study of speaker adaptation methods for HMM-based speech recognition", In ICSLP-1990, 145-148.