SIGN LANGUAGE ANALYSIS USING ARTIFICIAL INTELIGENCE

Abstract: This paper introduces “SOUL” Sign Language Understanding and Learning, a real-time sign language detection and translation system that aims to fill the communication gap between hearing-impaired people and non- signers. The proposed system uses a hybrid machine learning approach, which combines Convolutional Neural Networks (CNN) for feature extraction and Random Forest Classifiers for gesture recognition. The system uses computer vision, natural language processing (NLP), and text-to-speech (TTS) techniques to enable seamless bidirectional translation between sign language and spoken language. The proposed system has an accuracy of 94.2% in recognizing American Sign Language (ASL) alphabets and phrases, with a real-time processing speed of 18 FPS. Moreover, the system is capable of multilingual translations (English and Marathi), making it flexible for use in different linguistic settings.

Keywords: Sign Language Recognition, CNN, Random Forest, Real-Time Translation, TTS, Accessibility.