Support Technique for Real-Time Captionist to Use Speech Recognition Software

Abstract

Deaf or hard-of-hearing students who attend lessons at university acquire much less information than students with normal hearing. The captionist (transcriptionist) listens to the teacher and “re-speaks” (repeats exactly what is heard) what the teacher says. The clear and distinct speech that is repeated by the captionist is sent to the automatic speech recognition (ASR) software installed on a personal computer, which performs “speech-to-text conversion”. However, it is not easy to “re-speak”. The results of the research suggest that listening through headphones with superior sound-proofing features enables captionists to sustain re-speak ability compared with listening directly. These facts indicate that the recognition accuracy of existing ASR technologies could be maintained.