부산대학교 도서관

As there has been a tremendous development in “on-the-fly” computing technology, it becomes very essential to secure the system from unauthorized users and hackers. The biometrics verification system is one of the technological advancements in security measurements to restrict the unwanted users from accessing the systems. A behavioral biometric recognition and verification system performs the functions of reading unique biometric identity of an individual, such as fingerprint, iris, voice, and face, and authenticates the individual by comparing with the biometrics of the same kind in the database. As the above said behavioral biometric value can never be the same for two persons, the biometric verification and authentication system is highly accurate and reliable. The voice recognition and authentication system is one of the familiar and upcoming system that is taken for discussion in the chapter. To add more accuracy and intelligence flavor to the biometric especially voice-based verification and authentication, deep leaning is adopted as a pre-verification filter to separate out and remove bad or malicious individual from accessing the system. Deep learning puts core architectural building blocks together to build a new voice-based verification specific architecture. Deep learning allows the system to train itself again and again with test samples and makes the behavioral voice-based verification and authentication system (BVVAS) to be more secure and achieves high level of accuracy by reducing false identification. The deep learning supported BVVAS appears to be obviously fit for handling the everrising biometric identification problems ranging from mobile device to high-end security systems authentication.