Speaker verification (authentication using voice biometrics) systems are already in use by banks and other financial institutions. Voice recognition systems are becoming more and more common in systems ranging from home IoT, cars, mobiles and desktop operating systems. So speaker verification appears to be a potential method of interacting with these devices in the future.
The talk will present a methodology for testing the security of speaker verification systems including developing tools and exploring the types of attacks possible against voice biometric authentication systems. The results of attacks against a variety of speaker verification APIs will be discussed along with live demos and a tool release.