CineAD. A system for automated audio description script generation for the visually impaired

Audio description (AD) is an assistive technology that allows visually impaired people to access cinema and understand the story of a movie. Basically, the visual content of the story is told by way of using a voice, narrated during the flm gaps of silence. Nonetheless, this assistive technology is not widely used, due to several factors, among them the high cost and time involved in creating such audio descriptions. Towards solving this problem, this work proposes a solution that automatically generates AD scripts for recorded audiovisual content, named CineAD. This solution detects the breaks in the spoken lines in the video receiving the AD and generates these descriptions from the original script and subtitles. Alternatively, the solution can be incorporated into a speech synthesizer or used by an audio description narrator to generate the audio that contains the descriptions. To evaluate the proposed solution, qualitative tests with visually impaired users and audio description narrators are conducted. The results show that the proposed solution can generate descriptions of the most important events in the videos, and therefore, can help to reduce the barriers in accessing video faced by visually impaired, when the script and subtitles are available.