Interactive Storytelling with Gaze-Responsive Subtitles

The paper describes an eye-tracking framework for offline analysis of and real-time interaction with gaze-responsive subtitled media. The eventual goal is to introduce and to evaluate gaze-responsive subtitles, which afford pausing of video when reading subtitles. Initial modes of interaction include: look-to-read and look-to-release. The former pauses video as long as gaze is detected over subtitles, the latter pauses video until gaze falls on subtitles. To avoid disrupted perception of media content, an additional ambient soundtrack matched to the general content of video is proposed. Note that this is potentially revolutionary as it would require an entirely novel approach to film direction. Just as Audio Description is now included in most modern films, ambient sound would also be needed to fill in brief temporal gaps when the user’s visual attention is directed toward subtitles. Concomitantly, the eye-tracking framework fosters quantitative analysis of attention to audiovisual content apart from qualitative evaluation on which most of subtitling standardization is based.