• Balaga Pravallika
• Pappala Mohan Rao

This project introduces a real-time Sign Language Recognition system designed to help individuals with speech and hearing impairments communicate effectively. It uses a custom-trained Convolutional Neural Network (CNN) to classify 26 static hand gestures (A-Z) via webcam input. The gestures are preprocessed (grayscale, normalization, Gaussian blur) to enhance accuracy. The model achieved impressive accuracy levels—96% during training, 98% on the validation set, and 98% on the test set—leveraging the Adam optimization algorithm and softmax activation function [14][15]. During real-time execution, the system dynamically extracts the Region of Interest (ROI) to detect hand gestures and utilizes a prediction buffer to ensure stable and reliable recognition over time. Once identified, characters are sequentially combined into meaningful sentences, which are then vocalized through a speech engine powered by pyttsx3.Keyboard shortcuts enable user-friendly text editing. Unlike costly sensor-based alternatives, this vision-based system uses OpenCV, TensorFlow, and pyttsx3 in Python, making it both cost-effective and scalable. The project demonstrates a powerful, accessible solution in assistive technology for the speech- and hearing-impaired.

Convolutional Neural Network (CNN), Text-to-Speech (TTS), OpenCV, TensorFlow, pyttsx3, Region of Interest (ROI), Machine Learning, Vision-based System.

"Real-Time Sign Language Recognition Using Machine Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 7, page no.f606-f612, July-2025, Available :http://www.jetir.org/papers/JETIR2507564.pdf

"Real-Time Sign Language Recognition Using Machine Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 7, page no. ppf606-f612, July-2025, Available at : http://www.jetir.org/papers/JETIR2507564.pdf