Intelligent voice assistant for real-time translation between French and English

Abstract

In light of the rapid developments in the field of artificial intelligence, real-time voice translation technologies have become a pivotal tool for bridging linguistic and cultural gaps between individuals. This project aims to develop an intelligent voice assistant capable of real-time speech translation between French and English, while maintaining accuracy, con text, and meaning. The system is based on a Transformer architecture and combines tech niques such as Automatic Speech Recognition (ASR), Neural Machine Translation (NMT), and Text-to-Speech (TTS). Translation models have been trained on a custom dataset using f ine-tuning techniques to enhance performance in real-world use cases. The work also included a thorough evaluation of model performance using metrics such as accuracy and BLEU, which demonstrated promising results surpassing some common systems. This project highlights the advanced potential of integrating artificial intelligence in building effective and multilingual communication tools