SBIR-STTR Award

Proposal for development of neural network based machine translation engines integrated seamlessly with advanced front-end speech to text modules advanced back-end speech generation modules, over a network.

Benefits:
Cross-language communications applications for use in law enforcement, clinical medicine, tourism and emergency humanitarian relief, as well as educational and business applications over the internet

TheSmallWorld.com, Inc. is proposing to conduct R&D with the objective of producing a working prototype that will connect a customized online linguistic data center for unique speech-to-speech and text-to-text natural language translation applications to a mobile communications system. The main focus of our research will provide specific solutions for a variety of real-time translation applications for both military and commercial purposes. Understanding the limits of language translation technology and computational linguistics, we recognize that we must develop systems that incrementally reduce translation error rates, in part by recognizing and constraining subject domains and by customizing dictionaries that match these constraints. The practical application of speech recognition and speech generation to machine translation systems using wireless internet interfaces such as WAP portals on web-enabled cell phones, connected to LEO satellites, will involve development of middleware, (i.e. software systems architecture that bridges gaps between each component of the system).

Keywords:
Machine Translation Speech-To-Speech Speech Recognition Speech Synthesis Linguistic Data Center Translingual Communication Real-Time Translation Appli