Speech Recognition in Noisy Environments for Enhanced Tactical Performance

Recent advances in computer recognition of human speech have made it possible to perform reliable recognition of the content (words) in continuous speech, independent of the speaker. Using these new speech recognition tools together with algorithms which allow interpretation of the spoken messages, it's possible to dramatically reduce the human effort required to process spoken information, while improving information flow. We propose to use the expertise we have gained from developing our national-award winning VICTOR system and our SPRITE speech recognition and processing system, together with our proven expertise in signal processing and noise removal algorithms, to improve voice communications in high noise environments. We will design a speech recognition system with high accuracy in high noise environments that automatically responds to voice commands and sends this information to external systems, thus reducing required human effort, improving performance, and increasing the timeliness of information delivery. Goals include: 1.Designing a flexible, highly accurate speech recognition system that is usable in a high noise, multi-peopled environment; 2.Adapting and implementing noise reduction algorithms resulting in relatively noise-immune speech recognition; 3.Evaluating these noise reduction algorithms and the resulting speech recognition performance; 4.Demonstrating an early prototype system. BENEFITS: The most immediate applications are to improving soldier performance. User-friendly, flexible voice operated systems reduce operator training time and increase operator efficiency. Analogous systems also exist for the Navy, Air Force, Marine Corp., etc. Civilian applications include manufacturing (robotics and quality control databases), access control (voice command and logging systems), emergency dispatch (police, fire, etc.), and others.