Toward a dynamic framework for security evaluation of voice verification systems

Research results have shown that spoof attacks pose severe security threats to biometric verification systems. Nevertheless the literature lacks a comprehensive, flexible and dynamic framework for security analysis of biometric and especially voice based verification systems when spoof attacks are taken into picture. This paper aims at highlighting the vulnerabilities, classifying the threats and clarifying requirements for such a dynamic framework which transfers the vulnerability evaluation results into legacy knowledge in the security domain. The proposed design and architecture facilitates comparison of two systems as well as definition of the threats in an independent and isolated way.