Smart Glove for Sign Language Translation

Abstract

Communication is a fundamental right, yet individuals with hearing and speech impairments face significant barriers in daily interaction with a general population that does not understand sign language. Existing technology-based solutions, such as camera-based gesture recognition systems, are often expensive, computationally intensive, and dependent on environmental conditions. This paper proposes a wearable Gesture-Based Voice Communication System—implemented as a Smart Glove—that performs real-time translation of hand gestures into text and speech. The system integrates five flex sensors for per-finger bend detection, an MPU6050 six-axis inertial measurement unit for hand-orientation tracking, an Arduino Nano ATmega328P microcontroller for signal processing and gesture matching, an HC-05 Bluetooth module for wireless text transmission to a smartphone, and a DFPlayer Mini MP3 module for direct audio output. A threshold-based algorithm maps sensor combinations to a vocabulary of eight predefined phrases. Evaluation across 1,200 trials yielded an overall gesture recognition accuracy of 96.8 % with a mean end-to-end latency of 340 ms from gesture stabilisation to speech onset. A four-hour continuous-operation test recorded zero Bluetooth disconnections and zero playback failures. The system is portable, cost-effective, and independent of lighting or background conditions, providing a practical foundation for assistive communication technology