Speech Recognition with Hidden Markov Model and Multisensory Methods in Learning Applications Intended to Help Dyslexic Children

Abstract

Dyslexia disorders commonly occur in children. Dyslexia is a difficulty in recognizing letters and numbers. Helpful tools are needed to detect and treat dyslexic children so that early anticipation can be taken. So that later this disorder will no longer be suffered by the child when he is an adult. For this reason, information technology-based learning media was created in this study with the aim of helping therapy for children with dyslexia. This study utilizes the hidden Markov method for speech recognition and multisensory methods for learning media. In this study learning is divided into three levels, namely learning to recognize numbers, learning to recognize letters, and learning to understand words. Each level uses a pattern that is generally difficult to accept children with dyslexia disorders. During the testing process, level one and level three efficiency has reached 80%, while for level two it is still worth 60%.