A Rule Based System for Bangla Voice and Text to Bangla Sign Language Interpretation

摘要

It is very difficult in Bangladesh to make communication between Sign and Non-Sign people even though in the entire world because most of the Non-Sign people do not understand the Sign language. To create effective communication between Sign and Non-Sign people, we have proposed a rule-based system for Bangla Voice and text to Bangla sign language interpretation. Voice and Bangla Text are taken as input by the proposed method and this input text is parsed based on Bangla linguistics. After parsing, the system links and displays the signs with the corresponding trained animated Bangla Signs. The 2D animation for the model is trained with 14 animated Bangla Numeral Signs (0-9, Hundred, Thousand, Lakh, Core). The system is being evaluated using 100 different lengths of Bangla Composite Numeral. The system achieves the interpretation average accuracy of 96.03% for voice, 100% for Bangla Text as the input of Bangla Composite Numeral, and Bangla Sentences with a computation cost of 9.5ms per sign.