Published Paper Details:

: Individuals with speech and hearing impairments face persistent communication barriers, contributing to social isolation and limited accessibility. While American Sign Language (ASL) remains a critical communication tool, its visual nature restricts interactions with non-signers, underscoring the need for inclusive technological interventions. This research introduces a wearable sign language-to-text translation system that synergizes deep learning, multi-sensor fusion, and IoT to convert ASL gestures into real-time text. The device employs sensor-embedded gloves with five flex sensors per hand to measure finger flexion and an MPU-6050 IMU to capture hand orientation, acceleration, and rotation. These sensors collectively track intricate gesture dynamics, such as finger articulation and palm movement, essential for accurate ASL interpretation. An Arduino Nano microcontroller preprocesses and wirelessly transmits sensor data to a hybrid CNN-RNN deep learning model. The CNN extracts spatial features from sensor inputs, while the RNN deciphers temporal patterns in gesture sequences, enabling robust recognition of complex signs. A custom dataset, incorporating diverse signing speeds, hand sizes, and environmental variables, ensures adaptability across users. Edge computing on the microcontroller enables real-time processing, while IoT integration streams translated text to smartphones or displays instantaneously. This approach circumvents limitations of camera-based systems, such as lighting sensitivity and occlusion, through direct sensor-based gesture capture. The ergonomic, lightweight glove design prioritizes comfort for prolonged daily use. Experimental trials achieved 94.6% accuracy across 50 ASL phrases, validating the system's reliability. By transforming gestures into accessible text, this innovation empowers deaf and hard-of- hearing individuals to communicate seamlessly in educational, professional, and social settings, fostering inclusivity and advancing assistive technology