Development Of An Interactive Language Learning Application For Children

Language development in children is a sequential process that starts from cries and babbles to the production of words and phrases. It has been found that children learn to speak by mimicking sounds within their environment. Speech recognition and synthesis applications can be used to improve the language development process. However, due to the differences between speech features of children and adults, speech recognition systems trained on adult speech do not generalize well when tested on child speech. This project implements a web application that is aimed at improving language development in children. The web app is incorporated with a speech recognition system that works well with child speech. It is built using the transformer model which has shown remarkable results on the speech recognition problem. The model is trained using child speech to increase the accuracy when tested on child speech data. The system is designed to be easy to use, child-friendly and efficient..