Improving Speech Recognition to Assist Real - time Classroom Note Taking

Speech recognition systems have advanced to the point where they are a viable option for providing note taking assistance for deaf and hard of hearing students. College lectures, which frequently contain domain-specific or uncommon terminology, provide a challenge for these systems that typically rely on a dictionary of common words to guide recognition. This paper reports on a prototype text analysis software tool, and some general configuration techniques, that can improve the ability of an affordable and off-the-shelf speech recognition system to assist deaf and hard of hearing students with note taking in a college classroom setting. Some specific technology choices are discussed and the results of a preliminary evaluation of the text analysis tool are presented.