Um serviço web para geração automática de audiodescrição de vídeos digitais

Abstract

Audio description is an accessibility feature designed to bene t visually impaired people in the context of visual information comprehension. According to Szarkowska (2011), some of the major obstacles to the popularization and availability of audio description are the generally time consuming production and the high costs involved during the process of generating this content. In order to reduce these problems, the present work proposes the design and development of a RESTful web service for the automatic generation of audio description tracks of digital videos. The solution developed utilizes strategies of automatic generation of audio description script based on information extracted from the video and its script, besides using Text-To-Speech (TTS) tools in conjunction with synchronization techniques to generate audio description tracks. The results showed that the solution was able to considerably reduce the time of audio creation of audio description resources when compared to the traditional methodology. Additionally, considering that most of the videos currently available do not have audio description, the proposal can be considered a viable alternative when human audio description is not available or is not feasible.