DyslexiEase: Enhancing Dyslexic Learning Through Deep Learning-Based Multimedia Content Generation

According to research, around 10-15 percent of students have some form of dyslexia. This neurological disorder necessitates addressing it at a younger age in order to deal with its consequences quickly and effectively. This paper presents DyslexiEase, an innovative, child-friendly multimedia content generation tool designed to enhance classroom experiences for dyslexic students by incorporating a multimedia learning approach. By leveraging generative models, DyslexiEase transforms textual prompts into multimedia learning aids. Subject extraction is performed to identify key concepts, which are used by a diffusion model to generate reference images. Simultaneously, a self-trained Generative Adversarial Network (GAN) on the CleanVid15M dataset produces contextual videos, while Tacotron2 generates high-quality speech narrations. Additionally, a Conditional Variational Autoencoder (CVAE) creates captioned images using the OpenDyslexic font, with commonly misread letters highlighted for improved readability. The tool’s intuitive interface helps educators tailor content to students’ needs, promoting a more inclusive and engaging learning experience.