Language Based Speech Recognition Module

The implementation of a natural language based speech recognition module requires a significant processing resource to operate in real-time. This exploratory development program will exploit the power of a parallel digital signal processor technology to save the expense and size of a super computer solution. The significant challenge is to match the proposed algorithms and software operating system with the available digital Signal Processor (DSP) architecture to efficiently implement a language based speech recognizer in a microcomputer environment. The results will be a practical implementation in a small add-on parallel DSP/microcomputer module that interfaces to any standard computer configuration and permits real-time spoken language communications between the operator and the computer hardware and software. It will permit easy operator control of all computerized systems, from desktop to factory automation. Natural speech interface with automated systems represents a flexible and very high level means of communication with computers which will reduce training time, decrease operator response time, increase operator efficiency, and reduce the required number of workers or crew size. Synthesis technology will further ease the recognizer understanding giving the computer the ability to enter into a dialog with the operator.