Bayes factor scoring of GMMs for speaker verification

摘要

This paper implements and assesses the Bayes factor as a replacement verification criterion to the likelihood-ratio test in the context of GMM-based speaker verification. An advantage of the Bayesian method is that model parameters are considered random variables, allowing for the incorporation of prior information and uncertainty of parameter estimates into the scoring process. A novel development of Bayes factors for GMMs is presented based on incremental adaptation that is well-suited to inclusion in existing state-of-the-art GMM-UBM systems. Experiments on the 1999 NIST Speaker Recognition Evaluation corpus demonstrate improved performance over expected log-likelihood ratio scoring particularly when combined with the feature mapping technique.