A robust speaker verification biometric

The need for simple, ubiquitous security procedures to verify the identity of authorised system users is assuming increasing importance in the expansion and exploitation of new technology. High performance Biometric Verification methods offer important tools for this purpose, enhancing the security, reliability and integrity of transactions conducted electronically. This paper describes TESPAR/FANN, a new digital data/artificial neural network combination which is proving highly effective in the Speaker Verification arena, and in other key non-speech applications.