The proposed innovative development and demonstration of a pocketable voice-to-voice language translator is achievable because it uses LSIs current PC-based, speaker-independent, two- way voice-to-voice translation software as a foundation for the effort. The development will involve six technical tasks; the first four are planned for the initial Phase I effort, and the last two for a Phase I Option. Tasks 1 and 2 involve extension and enhancement of LSIs current voice translation software using VoiceXML and Java to provide platform independence and a capability for remote activation. Task 3 is dedicated to dialog development/extension for a selected military application, and 4 is a test of the software developed under Tasks 1-3. Task 5 is a proof of concept demonstration for the standalone capability, and 6 involves experimentation with cellular, PDA, and hybrid devices to determine the optimal platform for both local standalone and remote capabilities. Motorola, which has expressed interest in the project, can assist LSI in testing the remote translation software, and in evaluating platforms for local and remote capability. The commercialization strategy for this development will follow LSIs strategy for commercializing our current voice- to-voice translation software, originally ported to the PC environment under a previous DARPA/Air Force effort.
Benefits: The pocketable voice-to-voice translator will benefit LSIs current law enforcement clients, as well as medical and paramedic service providers, retail financial service providers (banks, loan companies, real estate companies, insurance companies), organizations with multinational operations, business travelers and tourists.
Keywords: voice-to-voice, language translation, speech recognition, speech synthesis, VoiceXML, language training
