Real Time Sign Language Conversation Using AI

Abstract

This project introduces a real-time Indian Sign Language (ISL) recognition and translation system to improve communication between hearing-impaired and non-signing individuals. Using OpenCV and MediaPipe, the system accurately detects and tracks hand landmarks to recognize gestures made with one or both hands. Each gesture is mapped to a predefined ISL phrase, which is then converted into text and spoken output using a non-blocking text-to-speech engine (pyttsx3) for smooth and simultaneous operation. The system also supports text-to-sign translation, displaying pre-recorded ISL videos with synchronized speech for interactive learning. Developed entirely in Python, the model provides real-time performance and requires minimal hardware. It can be applied in educational, public, and assistive communication systems. Overall, the proposed system offers an efficient, affordable, and inclusive solution that enhances accessibility and bridges communication barriers for individuals with hearing and speech disabilities.