AI-generated tactile graphics for visually impaired children: A usability study of a multimodal educational product

Approximately 70 million children aged 0–14 worldwide have visual impairments, limiting their language acquisition and image recognition development due to a lack of visual input. Human-computer interaction technologies provide an opportunity to learn Braille and images through touch and auditory stimuli, replacing traditional visual input elements. However, effectively integrating these sensory inputs remains a challenge. To address this, we developed "TaleVision," a product that uses large language models to instantly generate touchable images and sound effects, allowing visually impaired children to better understand the connection between Braille and images. We recruited five sighted children and 14 children with congenital visual impairments to test our product. Almost all participants reported high satisfaction and rated the user experience highly. Additionally, we also observed how visually impaired children and teachers interacted with our products to improve teaching and learning. Based on the results, we discussed best design practices, process recommendations, and potential future solutions.