Biometric voice authentication auto-evaluation system

Authentication that is based on Biometrics traits is more secure for many applications like cryptosystems, therefore; the template of stored biometric members of the biometric authentication system is a dangerous issue, while it can be stolen or breached. We produce in this paper an automatic evaluator depending on the biometric performance measures FAR, FRR and EER by simulating how noise in the transmission and background environment may have effect on the voice signals in the biometric template. Many types of research focused on just the rate of the FAR, FRR and ERR of the authentication system that can produce. This paper introduce a new view of how to see the database template from the statistical side depending on biometric performance measures. We test our new view by producing an evaluator system and perform it on the TIMIT speech database by using AWGN to simulate white noise in the transmission and background environment with deferent SNR levels. We found -from the empirical method- promising results and we hope that this paper spots some light on how to evaluate biometric voice template automatically and providing a report about expected accuracy and security.