Sign Recognition And Voice Conversion Device For Dumb

This project makes a small, low-cost device that helps people who can’t speak. A tiny camera on light glasses watches the user’s hand signs and sends those images to a small processor. The processor uses a fast, simple neural network to recognize 60 common signs—letters, words, and short phrases—with over 95% accuracy. Once a sign is understood, the device speaks it out loud through a built-in speaker in under a quarter of a second. Two easy buttons let the user switch modes, change volume, or pause. The device runs for about eight hours on one battery charge. Early tests with eight users showed it is accurate, quick, and comfortable to wear. Future updates will add full sentences, regional sign versions, and a phone app to add new words.