Digital TV Based Sign Language Avatar

This paper details a methodology to enable inclu-sive content for deaf or hard of hearing users on multimedia systems and TV appliances, improving accessibility in digital TV environment. With this methodology, the worldwide broad-casters would be able to add an additional elementary stream with bitstream data representing the animation content in sign language expressions (also known as glosses), to be used by the receiver device to animate a 3D graphical humanoid model (Avatar). The method reuses existent digital TV transport layer structures to wrap said bitstream data and include it on digital TV stream, synchronized with audio and video (A/V) media flows. For technical feasibility analysis and transmission tests were used a Brazilian sign language translation and animation framework called VLibras. Even though VLibras is based on Unity, this methodology can be applied to any graphical engine that can be embedded on a receiver device by implementing custom signalization tags inside digital TV Program Specific Information (PSI) tables. The steps involved, from content signalization, data insertion (in broadcaster side), to reception and decoding on receiver side are deeply discussed in this work. In this research scope we performed experiments in a real digital TV environment, and its results subsidize the technical feasibility of deploying this methodology in any digital TV environment.