Deep Learning-Based Speech Synthesis for Enhancing the Learning Ability of Individuals with Dyslexia

This paper explores dyslexia recognition and expressive speech synthesis. Dyslexia, a neurological disorder, impairs reading and writing but not intelligence. Early diagnosis using machine learning is crucial. Reviewing 19 articles, the paper addresses challenges in data collection, preprocessing, and model evaluation for dyslexia prediction. It also examines the shift from conventional to deep learning models in speech synthesis, focusing on generating expressive, human-like speech. Additionally, the paper proposes using generative neural networks to produce audio of basic Hindi words to aid dyslexic children. This work provides insights and directions for future research in both dyslexia recognition and expressive speech synthesis.