rjohnso9 | tboult uccs.edu

wscheirer fas.harvard.edu This paper examines a novel security model for voice biometrics that decomposes the overall problem into bits of “biometric identity security, ” bits of “knowledge secu-rity, ” and bits of “traditional encryption security”. This is the first paper to examine balancing security gained from text-dependent and text-independent voice biometrics under this new model. Our formulation allows for text-dependent voice biometrics to address both what you know and what you are. A text-independent component is added to de-feat replay attacks. The paper experimentally examines an extension of the recently introduced Vaulted Voice Verifi-cation protocol and the security tradeoffs of adding these elements. We show that by mixing text-dependent with text-independent voice verification and by expanding the challenge-response protocol, Vaulted Voice Verification can preserve privacy while addressing the issues with voice as a remote/mobile biometric identifier. We show the combi-nation of text-dependent and text-independent biometrics in the enhanced Vaulted Voice Verification approach not only adds privacy, but also increases accuracy over the standard GMM model. The resulting model supports both authenti-cation and key release with the matching taking place client side, where a mobile device may be used. 1