ARCHIVES

Abstract: The Multilingual Voice Assistant will be a natural, Hindi, Tamil, and English multilingual interface, with deep neural networks (DNNs) for speech recognition and TTS synthesis that maintain linguistic properties. Despite the fact that it would translate spoken language and generate very natural answers, it will employ NLP techniques to interpret the intent of questions asked by the user, identify intents, and ignore irrelevant information. An RL algorithm is utilized for continuous adaptation and enhancement of interaction strategies by the assistant on receiving feedback from the user. Hindi and Tamil have unique sounds and grammatical structures, and the assistant will be trained for these sounds and structures to ensure meaningful reply creation and smooth communication. This project advances voice assistant technology in providing the same user-friendly platform to multilingual users, thus making the technologies much more accessible for speakers of Hindi and Tamil.