Matavan machine translation approach for text to sign language avatar animation

Abstract

Sign language is a spatio-gestural visual language, which is the primary native language for hearing impaired community. The modality of exchanging information and expressing individual idea in sign language is through signs gestures. Sign gesturing follows a non linear pattern and are limited in number. However sign language is not a universal language. Every country or community has its own dialect sign language. There are various sign languages used by hearing impaired people such as American Sign Language (ASL), British Sign Language (BSL) and Indian Sign Language (ISL) based on their topographical region. One such variant of ISL is Tamil sign Language (TSL) which is used in daily communication by Tamil hearing impaired community in the region of Tamilnadu. There exists a communication gap between hearing impaired people and Tamil speaking people of Tamilnadu. To bridge the gap between them a translation system that automatically translates text to sign language with proper grammar rules is needed. As-state-of art, existing sign language translation systems are domain specific, less effective, adopt a shallow transfer approach and not able to portray the gestures properly. Therefore there is need for a system that automatically and dynamically translates text to sign language gestures. The primary aim of this research work is to develop an automatic translation system for translating text to sign language avatar animation sequence. newline