A RNN based self-learning audio generating chatbot for French language learners

Abstract

A chatbot is an artificial intelligence-based virtual conversational agent that is one of the most potential technologies in the modern era. There are so many sectors where chatbots are providing outstanding performance. This technology has the ability to be an assistant for those who are trying to learn a new language. That is why we have done research to make a chatbot based on an RNN algorithm in order to help people learn French. Although there are many online platforms available to learn French, many of these are not sufficient enough to learn a language as it is mandatory to have a conversational partner to practice a language. Without practicing a language with anyone, it is difficult for a person to keep memorizing a lot of vocabulary. French pronunciation rules are not the same as English. Therefore, we have added a feature for sending text along with audio messages and getting replies in French audio alongside text. Because of having these features, users will not only know the French vocabulary and grammatical rules but can also be introduced to French pronunciation systems and learn to speak French with a native French accent. The self-learning feature assists this chatbot to increase its capability of giving correct responses to the users day by day by having continuous conversations and removes the problem of collecting so much data to make this kind of chatbot. Also the paraphrasing feature will help users to get different replies to a particular question at different times, enriching their vocabulary and familiarizing them with the variety of replies to a particular question.