Sign Language Interpreter through Animation

Abstract

Introduction: Sign language is a natural mode of Communication for those with hearing or speech impairments. Despite the fact that there are many systems for deciphering and identifying sign language, little is known about text-to-sign language translation systems because there aren't many linguistic corpora available.

Objectives: This study suggests a translation framework that translates English sentences into the grammar of Indian Sign Language (ISL). To ensure compatibility with ISL grammar, which does not allow inflections, the system uses a parser module to create phrase structure grammar representations. This is followed by stopword removal and stemming. Every processed word has a corresponding video in the ISL dictionary, and terms that are not in the vocabulary are replaced with synonyms

Methods: The suggested method uses ISL-specific grammatical rules to produce translations that are both semantically and syntactically accurate, in contrast to current methods that carry out direct word-to-sign mapping. By facilitating more organic interactions between non-signers and the hearing impaired, this framework seeks to close communication gaps.

Results: When compared to baseline systems, preliminary evaluation shows improved translation accuracy and comprehension.

Conclusions: Rather than representing individual words with GIFs or images, the model successfully transforms the entire input text into a single visual representation, enhancing the model's realism and liveliness. However, there is a need for further expansion of the ISL lexicon, which presents an opportunity for future development in this area.