Speech Recognition is a captivating research field of Artificial Intelligence. Starting with the beginning of the last century, the first applications of speech recognition were toys reacting to vowels uttered by children. New applications are dictation systems, dialog systems, interfaces offering improved accessibility for handicapped persons, and search in databases of speech files. Up to our days, the main civil application of speech recognition seems to remain with the toy industry. Here, toys are expected to spot certain special key-words (like names and commands) and to react. The inherent lack of robustness of speech recognition can be turned by a toy into a quality. Despite its relatively long history in artificial intelligence, keyword spotting still offer possibilities of improvement. This talk will introduce a simple and very efficient technique for solving this problem.
The speaker is Assistant Professor at the Florida Institute of Technology. He has started to work in the area of speech recognition in 1998 while he was associated with the Swiss Federal Institute of Technology. His main contributions are in the area of keyword spotting, where he has developed a suite of efficient algorithms for maximizing existing and new recognition criteria. The speaker got his PhD in 2002 from the Swiss Federal Institute of Technology at Lausanne with a dissertation on "Asynchronously Solving distributed problems with privacy requirements", and the BS in 1997 from the Technical University of Cluj-Napoca. He is teaching courses of "Artificial Intelligence", "Speech Recognition" and "Cryptography and Multi-party Computations".
Sprecher: Dr. Marius Silaghi Wann: Dienstag, 11. Mai 2004, 16:00 Uhr (s.t.) Wo: E 2.69, Universität Klagenfurt