STTR Phase I: Providing Access to Science and Math for Disabled Students Using MathSpeak

This Small Business Technology Transfer Research (STTR) Phase I research project investigates the improvement of the quality of the synthetic speech rendering generated by the conversion mechanism and to subsequently demonstrate Phase II commercialization feasibility via efficacy tests with targeted consumer groups. The first objective will be to remove distortions and mechanical sounding expressions. Observations suggest that the distortions arise from the lack of logical insertion of pauses (i.e., based on a model of real speech) between elements in the synthetic speech stream. This objective will be accomplished by developing an algorithm based on recordings of math educators speaking mathematical expressions that will be incorporated into the synthetic speech conversion mechanism. The second objective will involve efficacy testing for the reduction of distortions and mechanical sounding expressions. Distortion reduction testing will require participants to listen expressions and to exactly (verbatim) report the expressions they heard. The third objective will involve an efficacy test to test the capacity of the pause algorithm to enhance disambiguation.Broader impacts include the following: an overriding goal is to increase the accessibility to science, technology, engineering, and mathematical fields by under-represented groups. This goal is consistent with the objectives of the NSF and the Department of Educations incentive to further develop and implement the National Instructional Accessibility Standard (NIMAS), which would facilitate an increase in accessibility via the development of flexible alternatives to print as a primary objective. Additionally, the project advances discovery and understanding not only of the specifics aspects of the wording of spoken language that contribute to misinterpretations but also investigates the more subtle nuances of the non-verbal information such as the intervals between word and their influences on accurate acquisition of information. These findings have educational implications for the optimization of learning strategies and models for teaching and training. Further, the basic protocol for algorithm development has potential paradigmatic significance for development and improvement of synthetic speech across multiple fields that synthetic speech such as artificial intelligence, augmentative and alternative communication, and computer and communication sciences.