Improving Real-Time Captioning Experiences for Deaf and Hard of Hearing Students

Abstract

We take a qualitative approach to understanding deaf and hard of hearing (DHH) students' experiences with real-time captioning as an access technology in mainstream university classrooms. We consider both existing human-based captioning as well as new machine-based solutions that use automatic speech recognition (ASR). We employed a variety of qualitative research methods to gather data about students' captioning experiences including in-class observations, interviews, diary studies, and usability evaluations. We also conducted a co-design workshop with 8 stakeholders after our initial research findings. Our results show that accuracy and reliability of the technology are still the most important issues across captioning solutions. However, we additionally found that current captioning solutions tend to limit students' autonomy in the classroom and present a variety of user experience shortcomings, such as complex setups, poor feedback and limited control over caption presentation. Based on these findings, we propose design requirements and recommend features for real-time captioning in mainstream classrooms.