Voice To Sign Language Conversion

ABSTRACT True incapacity can be seen as the inability to speak, where individuals with speech impairments struggle to communicate verbally or through hearing. To bridge this gap, many rely on sign language, a visual method of communication that uses hand gestures. Although sign language has become more widespread, interaction between those who sign and those who don't can still pose challenges. As communication has grown to be an essential part of daily life, sign language serves as a crucial tool for those with speech and hearing difficulties. Recent advances in computer vision and deep learning have significantly enhanced the ability to recognize gestures and movements. While American Sign Language (ASL) has been thoroughly researched, Indian Sign Language (ISL) remains underexplored. Our proposed approach focuses on recognizing 4972 static hand gestures representing 24 English alphabets (excluding J and Z) in ISL. The project aims to build a deep learning-based system that translates these gestures into text using the "Google Text-to-Speech" API, thereby enabling better interaction between signers and non-signers. Using a dataset from Kaggle and a custom Convolutional Neural Network (CNN), our method achieved a 99% accuracy rate. Key Words: Convolutional Neural Network; Google text to speech API; Indian signing.