A comparison of Gaussian mixture and multiple binary classifier models for speaker verification

A Gaussian mixture model (GMM) is compared to a multiple binary classifier model (MBCM) in two speaker verification experiments conducted on telephone speech. The MBCM consists of 45 Moody-Darken radial basis function neural networks (MD-RBFNs) whose outputs are fused. Furthermore, the model is pruned in order to remove poorly performing MD-RBFNs. In the first experiment, true speakers and impostors are selected within the same dialectic region. The latter claim the identities of each of the former, in turn. The MBCM outperforms the GMM, both before and after pruning, by 21% and 62% respectively. The experiment is repeated, selecting impostors from outside the true speakers´ dialectic regions. In this case, the mean MBCM performance lags that of the GMM by 10% before pruning, but outstrips the latter by 16% following pruning