One of the greatest challenges facing American educators as we enter the next century involves the need to provide mainstream classroom instruction simultaneously to groups of students comprised of diverse heterogeneous cultures and language backgrounds. Challenges include both cost and technical components, since it is not often feasible to hire multi-lingual teachers or aides for every class and every language in every classroom. There are also many complications involved in attempting to deliver Multi-Lingual Classroom (MLC) instruction that satisfies all social and financial objectives and still preserves quality of education for all students across all academic levels, speeds, and learning styles. Several promising technological solutions to these problems involve new speech recognition API (SAPI) application protocol interface standards for PC software as a means of automatically translating continuous speech of both the teacher and students. However, several significant system architecture and cost problems exist related to its delivery and performance on multimedia PC systems in classroom settings. Such systems must be affordable and powerful enough to handle robust vocabularies or syntax, and be contextually re-configurable according to course content. An adaptive system that meets basic requirements will be provided based on Distributed Component Object Model (DCOM) standards, and parallel virtual machine (PVM) PC architectures.Summary of anticipated results and implications:This project will result in demonstration proof-of-concept of the use of AI speech recognition engine software in multi-lingual classrooms, and lead to products that support simultaneous universal translation.
Potential Commercial Applications: This research will lead to numerous proprietary commercial products and services for universal translation in group settings including education, business communication, diplomacy and the travel industry.
