An investigation into the use of Bayesian learning of the parameters of a multivariate Gaussian mixture density has been carried but. In a continuous density hidden Markov model (CDHMM) framework, Bayesian learning serves as a unified approach for parameter smoothing, speaker adaptation, speaker clustering and corrective training. The goal is to enhance model robustness in a CDHMM-based speech recognition system so as to improve performance. Our approach is to use Bayesian learning to incorporate prior knowledge into the training process in the form of prior densities of the HMM parameters. The theoretical basis for this procedure is presented and results applying it to HMM parameter smoothing, speaker adaptation, speaker clustering, and corrective training are given. The following word error reductions were observed on the DARPA RM task: 10% with HMM parameter smoothing, 31% for speaker adaptation with 2 minutes of speaker specific training data, and 15% with sex-dependent modeling.
Bibliographic reference. Gauvain, Jean-Luc / Lee, Chin-Hui (1991): "Bayesian learning for hidden Markov model with Gaussian mixture state observation densities", In EUROSPEECH-1991, 939-942.