Towards human-centered optimization of mobile sign language video communication
Jessica J. Tran · ACM SIGACCESS Accessibility and Computing · 2013
The mainstream adoption of mobile video communication, especially among deaf and hard-of-hearing people, is heavily reliant on cellular network capacity. Video compression lowers the rate at which video content is transmitted; however, intelligibility may be sacrificed. Currently, there is not a standard method to evaluate video intelligibility, or a good communication model on which to base evaluation. I am developing a better theoretical model, the Human Signal Intelligibility (HSI) , to evaluate intelligibility of lowered video quality for the purpose of reducing bandwidth consumption and extending cell phone battery duration. The goal of my dissertation is to advance mobile sign language video communication so it does not rely on higher cellular network bandwidth capacities. I will conduct this work by (1) identifying the components in the HSI model that make up intelligibility of a communication signal and separating those from the comprehensibility of a communication signal; and (2) using this model to identify how low video quality can get before the intelligibility of video content is sacrificed. Thus far, I have evaluated the use of mobile video communication among deaf and hard-of-hearing teenagers; developed two new power-saving algorithms; and quantified battery savings and evaluated user perception when those algorithms are applied.