Language Translator with Machine Learning (English-French)

Abstract

In today’s globalized world, the need for efficient language translation tools has become critical. This project aims to develop a “Language Translation System” by integrating GNMT (Google Neural Machine Translation) Translation API and a user-friendly Streamlit interface. The system leverages advanced deep learning models trained on extensive datasets to provide accurate and context-aware translations. By focusing on accessibility and precision, the project seeks to bridge linguistic barriers and facilitate seamless communication. Users can input English text, select the desired target language, and receive real-time translations, ensuring ease of use for all.

Introduction

Overview

In a globalized world, language translation is essential for bridging communication gaps across cultures and regions. Manual translation is often slow and error-prone, highlighting the need for automated, accurate, and fast solutions.

This project introduces a cloud-based neural machine translation system that leverages Google Neural Machine Translation (GNMT) to translate English to French using a Streamlit-powered web interface. The system offers a seamless, user-friendly experience with real-time translation and high accuracy.

Key Features & Objectives

• Purpose: Enable efficient, high-quality English-to-French translation.

• Tech Stack:

  • Frontend: Streamlit (Python)

  • Backend: Google Cloud Translation API using AutoML-trained model

  • Hosting: Google Cloud Platform (optionally Amazon EC2)

  • Security: Service account authentication & HTTPS

• Performance:

  • Trained on 79,625 sentence pairs

  • Achieved a BLEU score of 51.428, indicating strong translation accuracy

  • No local storage, ensuring user privacy

System Architecture

1. User Interface (UI):

   • Built with Streamlit as a single-page app

   • Accepts English text input and displays translated French output

2. Application Layer (Backend):

   • Handles input, communicates with Google API, processes and displays output

3. Data Layer:

   • Translation happens entirely in the cloud using Google’s pre-trained NMT model

4. Security & Integration:

   • Uses Google Cloud service accounts and encrypted communication

   • Integrates HTML encoding/decoding libraries

5. Deployment:

   • Hosted on GCP with options for horizontal/vertical scaling

   • Supports CI/CD (e.g., GitHub Actions), Docker, and Google Cloud Logging

Translation Workflow

1. User enters English text

2. Selects French as the target language

3. Text is sent to GNMT for real-time translation

4. Output is displayed back on the same interface

Translation Algorithm

• Uses Seq2Seq (Sequence-to-Sequence) GNMT model with attention mechanism:

  • Encoder: Converts English text to numerical representation

  • Decoder: Converts this into French

  • Uses ReLU activation and Softmax for word prediction

  • Translated text is post-processed and displayed in UI

Key Processes

• Input Validation: Ensures user text is non-empty

• Google Cloud Authentication: Via JSON key file

• Text Cleaning & Encoding: Prepares input for translation

• Translation Request: Sends input to GNMT via API

• Output Rendering: HTML-decoded and shown on frontend

Results

• High BLEU score (51.428) shows effective translation

• User interface is clean and intuitive

• The system ensures fast, secure, and accurate translations

• Future plans include adding Spanish and German, translation history, and further model fine-tuning

Conclusion

In summary, this project successfully develops a web interface using Python through which the GNMT from Google Cloud can be integrated into a real-time translator. The real-time translations of the application are made possible through the Google Cloud Translation API, and they are as accurate as they are context-related from English to French. The application is built using Streamlit, which provides a friendly experience for users interfacing with the application while it processes their translation requests efficiently. With robust error handling, the system is able to show increased reliability and stability. The modular design enables scalability by keeping it open to future enhancement possibilities such as other languages or speech-based features. It thus showcases together the practical application of artificial intelligence in minimizing language barriers for greater global communication through accessible and reliable technology.

References

[1] Somya Singh, Ishu Rattan, Darshan Kaur. \"Language Translation Using Machine Learning \" International Journal of Computer Applications, ISSN: 2320-2882, Volume 12, Issue 4 April 2024. [2] Srivani EN, Bhavana. \"Language Translation Platform on Machine Learning,\" International Journal of Research Publication and Reviews, ISSN 2582-7421, Vol (5), Issue (5), May (2024), Pages 3280-3283. [3] Geetha, K., Priyakumari, M. \"Machine Learning Based On Multilingual Translation.\" International Journal of New Innovations in Engineering and Technology, ISSN: 2319-6319l, Vol. 24, Issue 1, Mar. 2024, pp. 1421-1426 [4] Reballiwar L. V., Yergude S. B., Urade V. M., Birewar S. R., Karmarkar B. \"Language Translation Using Machine Learning,\" International Journal of Advanced Research in Science, Communication and Technology (IJARSCT), ISSN (Online) 2581-9429, Volume 3, Issue 1, December 2023, Pages 297-300. [5] Kharat, V., Chaudhari, U., Kesarwani, K., \"Regional Language Translator Using Neural Machine Translation,\" International Journal of Current Engineering and Scientific Research (IJCESR), ISSN (PRINT): 2393-8374 (ONLINE): 2394-0697 , Vol. 9, Issue 5, 2022.

Copyright

Copyright © 2025 Srisailanath , Nikhil P R, P Venkata Rakesh, Paniraj T, Polishetty Laxmi Sanjana. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.