SIGN LANGUAGE TRANSLATOR FOR SPEECH IMPAIRED PEOPLE

Sign language is a critical means of communication for the deaf and hard-of-hearing community.Translating sign language to speech can bridge the communication gap, fostering inclusivity.This work explores the use of Convolutional Neural Networks (CNNs) to develop a system that converts sign language gestures into spoken words.The process begins with data collection, involving videos or images of individuals performing sign language gestures, which are then preprocessed through frame extraction, normalization, data augmentation, and label encoding.A CNN model is designed with multiple convolutional and pooling layers to extract features, followed by fully connected layers for classification.The model is trained on a preprocessed dataset, with evaluation ensuring accuracy and generalization to new data.For real-time applications, a webcam captures video frames that the trained model processes to recognize gestures instantly.To convert recognized gestures into speech, the system maps predicted gestures to corresponding text labels, utilizing the pyttsx3 text-to-speech (TTS) library to vocalize the output.This dual-stage process of recognition and vocalization facilitates seamless communication.The proposed approach demonstrates the feasibility of using deep learning for sign language recognition and real-time translation to speech.This technology holds promise for enhancing communication accessibility, providing an essential tool for the deaf and hard-of-hearing community, and fostering greater social integration.