University of St. Thomas

Interests:

Automatic Speech Recognition, Voice Interfacing, Usability Engineering of Voice Systems, Knowledge Bases, and Artificial Intelligence.
 

Career Highlights:

Dr. Lea studied electrical engineering, cybernetics, linguistics, and speech science, with the goals of applying the best interdisciplinary science to practical projects in speech recognition and voice interfacing with machines. Primary algorithm contributions have been in prosodic aspects of speech recognition, such as syllable location, phrase boundary detection, location of linguistic stresses, pause and clause boundary detection, and the application of all such prosodic cues in guiding phonological and syntactic parsing of spoken English. He developed rhythm constraints on dynamic programming for alignment of speech segments with expected pronunciations, and demonstrated improved phonetic, word-matching, and sentence recognition with his methods. Dr. Lea also has been involved in the development of non-acoustic (electromagnetic) sensors of the movements of human articulators when one speaks. Dr. Lea also conducted basic scientific studies of human perception of prosody, theory of recognition, and methods for assessing the value of voice interfaces.

Dr. Lea’s primary focus while at UST has been on the integration of voice interactions with practical applications software, for spoken human interactions with automotive, truck, and telematics systems in vehicles, telephony and web interactions, and handheld interface products. He offers hundreds of usability guidelines for assuring user-satisfaction with voice interfaces, has documented hundreds of VUI bloopers to avoid, offers a full set of adequacy measures for voice interfaces, conducts user and task analyses to decide what tasks should be done by voice, and designs and conducts end-user tests of voice interfaces.

Dr. Lea currently serves as President and chief research officer of Speech Science Institute, Inc., and consults for leading fortune five hundred companies on the design of voice interfaces for various markets.  From 2006 to 2008, he served as Director of Voice User Interface Design at United Healthcare, and was Lead Achitect on Voice Interface designs for UHG consumers, healthcare providers, and others who called the corporate interactive voice response systems.

Previously, he served as corporate officer of several voice companies, and taught at the University of Southern California, University of California at Santa Barbara, and Purdue University. He has also taught over 40 short courses on voice interfacing and speech recognition at leading companies, government agencies, and professional groups, and chaired international conferences in voice interfacing and software design. Dr. Lea is the editor or author of several books, and has published over 100 articles, keynote talks, and conference presentations. He is currently implementing web-and-CD-databases for interactively guiding integrators in the design of accurate and user-friendly voice interfaces.

GPS Courses:

Speech Recognition -- SEIS 756
Voice Interface Development -- SEIS 757
 
Academic History:

Ph.D. (three departments: English, Electrical Engineering, and Audiology and Speech Science), Purdue University;
SM (Linguistics) and EE degrees from Massachusetts Institute of Technology;
BSEE and MSEE, Montana State University