This is an automatic method to set-up the several parameters that define the behavior and performance of a typical speech recognition engine. Such parameters include weights and beam widths among others. The method is based on the definition of a merit function. Here, merit is understood as an intuitive notion of recognition performance based on both recognition accuracy and computation time. A convenient definition of merit based on fuzzy logic allows to use a gradient-based optimization procedure to define a convenient set-up for the recognizer with little human intervention.

Speaker Bio:

Gustavo Hernandez Abrego received his B.S. degree in Electronic and Electric Engineering from La Salle Universitity Mexico in 1995 and the Ph.D. degree in Telecommunications Engineering from the Polytechnic University of Catalonia in Barcelona, Spain in 2000. In July 2000, he joined Sony Electronics' Spoken Language Technology laboratory where he was involved in the research and development of the speech recognition technology for Sony products. Particularly for the Sony entertainment robots AIBO and QRIO. In April 2004, he joined the Sony Computer Entertainment of America R&D division to work on the development of man-machine interfaces for the Playstation platform.