ACOUSTIC-LABIAL SPEAKER VERIFICATION

This paper describes a multimodal approach for speaker verification. T he system consists of two classifiers, one using visual features, the other using acoustic features. A lip tracker is used to extract visual information from the speaking face which provides shape and intensity features. We describe an approach for normalizing and mapping differe nt modalities onto a common confidence interval. We also describe a no vel method for integrating the scores of multiple classifiers. Verific ation experiments are reported for the individual modalities and for t he combined classifier. The integrated system outperformed each sub-sy stem and reduced the false acceptance rate of the acoustic sub-system from 2.3% to 0.5%. (C) 1997 Elsevier Science B.V.