Tamil Text to Sign Language Avatar System Enhancing Accessibility for the Hearing Impaired

Communication barriers between Indian Sign Language (ISL) users and Tamil speakers create significant challenges for individuals with hearing and speech impairments. To address this, we propose a two-way translation model utilizing a Flaskbased CNN-LSTM architecture for real-time ISL-to-Tamil and Tamil-to-ISL translation. The system is trained on a curated dataset comprising commonly used words in both languages, ensuring accurate and context-aware translations. Through rigorous testing, the model achieved an accuracy of 83%, demonstrating its reliability in facilitating seamless communication. This project serves as a crucial step toward inclusivity, with applications in language learning, real-time interactions, and accessibility technologies.