HandsSpeak: A Real-Time Interactive System for Translating Makaton Gestures into English

This preprint presents HandsSpeak, a real-time interactive system for translating basic Makaton gestures into English using a webcam-based computer vision pipeline. The system combines hand landmark detection, gesture classification, and an accessible graphical interface to support inclusive communication for individuals with speech and language difficulties. The work explores the use of multimodal artificial intelligence for augmentative and alternative communication (AAC), with potential applications in education, therapy, and digital health contexts. The prototype demonstrates how gesture recognition technologies can reduce communication barriers between Makaton users and non-users in real-world environments. This manuscript is a research preprint and has not yet been peer-reviewed.