Machine Leaning Based Urdu Language Tutor for Primary School Students

Abstract

Our proposed system is an offline android application named as “Urdu mu-all-um”. The final design is demonstrated through various android mobiles. The application consists of writing and reading of Urdu alphabets and words as well as reading of counting, grammar, idioms, proverbs, phrases, vocabulary, dialogues and chat-bot in Urdu language. We designed reading module to teach pronunciation of Urdu alphabets, words, and counting by creating data sets which include images and media player chunks. Media player is recording of pronunciation for each alphabet, word, and counting. Writing module teaches the writing by tracing on the canvas that was developed through Android Studio. In addition to tracing, a sub-part for practice is added for evaluation of the user by creating a model on Jupyter using python in deep learning and tensor flow and then the model was deployed into Android Studio. The sections on grammar, idioms, proverbs, phrases, vocabulary and dialogues are designed by using data through web services. Urdu chat-bot is designed by using wit.ai framework and preact. The project provides different basics of the language that are imperative for learning Urdu at a single platform.