Task-Oriented, Naturally Elicited Speech (TONE) database for the Force Requirements Expert System, Hawaii (FRESH).

Abstract

The Defense Advanced Research Projects Agency (DARPA) Strategic Computing has a goal of developing a large-vocabulary, speaker-independent voice-recognition system for battle management and fleet readiness assessment. One of the primary testbeds for the recognition system will be the command and control operation of the classified database for the Force Requirements Expert System, Hawaii (FRESH). The Naval Ocean Systems Center (NOSC) has designed an unclassified database called Task-Oriented, Naturally Elicited Speech (TONE) which simulates the characteristics of FRESH on a smaller scale. This study assisted NOSC in developing a voice-recognition, man-machine interface that could be used with TONE and upgraded at a later date for FRESH. The study identified more than 600 words that are associated with command and control and provided NOSC with the three most common forms of syntax used by the participants.