Automatic 3D Animated Bangla Sign Language Gestures Generation from Bangla Text and Voice
Md. Ariful Islam, Md. Ariful Islam, Ahsanul Karim, Muhammad Aminur Rahaman, Mahmuda Rahman, Md. Parvez Hossain, Syed Ahsanul Kabir · 2022
Every human being has a fundamental need for communication since it allows them to learn, express their emotions, and share ideas. However, sign language users are unable to hear or speak. They learn a variety of hand signals known as Sign Language (SL), which they use to communicate with themselves and others. Non-sign language users, on the other hand, have a difficult time understanding sign language. We have proposed a real-time bidirectional 3D Bangla Sign Language (BdSL) translator that is more dynamic and real-time than the traditional system for the simplest way to communicate. To facilitate effective communication between Sign and Unsign people, we have introduced an Automatic Animated BdSL Gesture Generation from Bangla Text and Voice. The recommended system takes input as Bangla Voice or Bangla Text and parses the output texts of corresponding inputs based on Signing Gesture Markup Language (SigML). Creating a bridge between Sign People and Normal People is the main purpose of this paper. A 3D animated Bangla sign gesture is generated dynamically based on the trained SigML. In this paper, the model is trained only for 13 Bangla Numeral gestures (0-9, Thousand, Lakh, and Crore) but the system can generate any 3D Animated Bangla Numeral gestures dynamically. 100 different lengths of Bangla Composite Numbers are being tested for the evaluation of the system. The system achieves an average interpretation accuracy of 97.80% for voice and 100% for Bangla Text input of Bangla Composite Numerals with a computation cost of 6.5 milliseconds per gesture.