Augmenting Captions with Emotional Cues: An AR Interface for Real-Time Accessible Communication

This paper introduces an augmented reality (AR) captioning framework designed to support Deaf and Hard of Hearing (DHH) learners in STEM classrooms by integrating non-verbal emotional cues into live transcriptions. Unlike conventional captioning systems that offer only plain text, our system fuses real-time speech recognition with affective and visual signal interpretation, including facial movements, gestures, and vocal tone, to produce emotionally enriched captions. These enhanced captions are rendered in an AR interface developed with Unity and provide contextual annotations such as speaker tone markers (e.g., "concerned") and gesture indicators (e.g., "nods"). The system leverages live camera and microphone input, processed through AI models to detect multimodal cues. Findings from preliminary evaluations suggest that this AR-based captioning approach significantly enhances comprehension and reduces cognitive effort compared to standard captions. Our work emphasizes the potential of immersive environments for inclusive, emotion-aware educational accessibility.