ES2ISL: An Advancement in Speech to Sign Language Translation using 3D Avatar Animator
Bhavinkumar Devendrabhai Patel, Harshit Balvantrai Patel, Manthan Ashok Khanvilkar, Nidhi Rajendrakumar Patel, Thangarajah Akilan
Abstract
This work proposes a model and an initial implementation of a robust system, which converts English Speech into Indian Sign Language (ES2ISL) animations. Such system may considerably enhance the lives of hearing-impaired people, especially in interaction and information exchange between concerned parties. The core purpose of the system is to bridge the communication gap between hearing-impaired people in India and others. It exploits and integrates the semantics of the Natural Language Processing (NLP), Google cloud speech recognizer API, and a predefined sign language database. The experimental results show that the proposed system outperforms existing models with an average accuracy of 77%. Hence, it overshadows the existing systems in terms of processing time by taking about 0.85s.