The proposed research addresses the problem of online authentication of a talkers identity using techniques borrowed from text-independent speaker identification. The proposed investigation will develop a basis for a Phase II, real-time system effort suitable for operational testing. The proposed algorithms build upon a number of prior developments by Sonetech and Lockheed-Sanders Inc., a proposed subcontractor, in optimizing feature sets and classification algorithms for real-world speaker identification systems. ANTICIPATED
Benefits: Authentication system suitable for use with high value STU III users, and for commercial and security voice authentication applications.