Third European Conference on Speech Communication and Technology

Berlin, Germany
September 22-25, 1993


Within Class Optimization of Cepstra for Speaker Recognition

J. Thompson, J. S. Mason

Department of Electrical & Electronic Engineering, University College of Swansea, Swansea, UK

By identifying highly speaker specific aspects of cepstral features, the task of speaker recognition can be potentially simplified. The identification process can be performed both phonetically and in the cepstral domain. The phonetic analysis aims to determine the temporal aspects of utterances which exhibit the highest degree of speaker specificity, while the cepstral analysis examines individual cepstra within these temporal divisions. This paper aims to complement work that has already been conducted on a phonetic basis, by performing analysis upon the individual cepstral coefficients within utterances.

Full Paper

Bibliographic reference.  Thompson, J. / Mason, J. S. (1993): "Within class optimization of cepstra for speaker recognition", In EUROSPEECH'93, 165-168.