PhD Dissertation Proposal by Rosiana NATALIE | The Design of a Collaborative Tool to Generate Scene Descriptions for People with Visual Impairments

The majority of online video content remains inaccessible for blind people due to the lack of audio descriptions despite the mandate from anti-discrimination related regulations for the provision of videos with audio descriptions (AD). Content creators have traditionally relied on professionals to author audio descriptions. How- ever, this gold standard may still not be accessible for casual content creators be- cause this method is still costly and has a long turnaround time.

To address this problem, I introduce four research threads for my Ph.D. dissertation. First, I developed and evaluated a mixed-ability collaboration authoring tool, called ViScene to understand the efficacy of collaborative methods in generating good-quality audio descriptions. Second, we extend ViScene so that it supports authors by automatically generating feedback on AD qualities. Third, I propose to explore the feasibility of AD-post editing to generate high-quality ADs where a program automatically generates the initial AD and humans will edit the generated audio descriptions. Lastly, I propose to explore the design of an accessible video player for blind people that can smartly incorporate ADs. I believe these four research threads will offer ways for the content creator to easily include accessibility in the videos they produce and blind viewers to consume video content.

Rosiana NATALIE is a Ph.D. student in the School of Computing and Information Systems, Singapore Management University. She is working with Assistant Professor Kotaro HARA. Prior to joining the Ph.D. program, she received her Master of Science Degree from National Tsing Hua University, Taiwan, in 2018. Her research interest lies in the intersection of Accessibility, Collaborative work, and Human-Computer Interaction. Her current research focuses on exploring and designing a collaborative method that is cost- and time-effective to author scene descriptions for blind people.