TERI RESEARCH INC  Providing the Information Advantage...                     

PRODUCTS       DEFENSE APPLICATIONS      KIOSK-COMMERCIAL       AMERICAN DISABILITIES ACT     

Emerging speech technology has been sufficiently matured to support a natural language human-computer interface. Command and information gathering applications can now be speech enabled with dialog feedback that mimics the rhythm and context building typical of human-to-human conversations. Real-time speech interface performance is achievable using the TRI developed NLS System software on a singular micro-chip host processor.   Natural language dictation transcription, with spoken command editing, is now practical with the advances made in automated profile learning and context adaptive technology.

Research is being conducted in complex text generation methods, synonym substitution, wider context interpretation and sensing speaker intent and focus shifts, to enhance the human-computer interface performance, so the speaker perceives that the conversation is with a human instead of a machine. However, the current real-time processing performance must be preserved for a natural dialog exchange and speech prosody. Variable dialog initiative modes of control proved to be the most natural method to achieve mutual interpretation understanding with a minimum number of dialog utterances.

TRI is continually working on improving and enhancing this technology.