Web-Based French to Yoruba Text Translation using Neural Machine Translation

Abstract

Language translation is vital for communication among individuals who speak different languages. Neural Machine Translation (NMT) models have advanced automated translation, improving accuracy and efficiency. This research developed a web-based app using NMT models for translating English to French and Yoruba. The research addresses the need for precise real-time translation services to bridge language barriers. Existing tools often lack accurate translations for specific language pairs, like French to Yoruba. The app utilizes NMT models to tackle this challenge effectively. A diverse dataset of English, French, and Yoruba text pairs were collected and preprocessed for training. Separate NMT models are trained for English to French and English to Yoruba translations, involving tokenization and data splitting. The app was built using Flask, with NMT models integrated into the backend for real-time translations. Users can provide feedback, stored in a MySQL database for analysis. Model performance is evaluated using split ratios (70:30, 60:40, and 90:10) with three (3) algorithms. Results demonstrate NMT models' efficacy in accurately translating English to French and Yoruba. The research showcases NMT models' potential for accurate translations between English, French, and Yoruba, contributing to improved language translation tools and better cross-lingual communication