5th European Conference on Speech Communication and Technology

Rhodes, Greece
September 22-25, 1997

Bayesian Affine Transformation of HMM Parameters for Instantaneous and Supervised Adaptation in Telephone Speech Recognition

Jen-Tzung Chien (1), Hsiao-Chuan Wang (1), Chin-Hui Lee (2)

(1) Department of Electrical Engineering, National Tsing Hua University, Hsinchu, Taiwan (2) Multimedia Communications Research Lab, Bell Laboratories, Murray Hill, USA

This paper proposes a Bayesian affine transformation of hidden Markov model (HMM) parameters for reducing the acoustic mismatch problem in telephone speech recognition. Our purpose is to transform the existing HMM parameters into its new version of specific telephone environment using affine function so as to improve the recognition rate. The maximum a posteriori (MAP) estimation which merges the prior statistics into transformation is applied for estimating the transformation parameters. Experiments demonstrate that the proposed Bayesian affine transformation is effective for instantaneous adaptation and supervised adaptation in telephone speech recognition. Model transformation using MAP estimation performs better than that using maximum-likelihood (ML) estimation.

Full Paper

Bibliographic reference.  Chien, Jen-Tzung / Wang, Hsiao-Chuan / Lee, Chin-Hui (1997): "Bayesian affine transformation of HMM parameters for instantaneous and supervised adaptation in telephone speech recognition", In EUROSPEECH-1997, 2563-2566.