🌍 Build Local Language Translator - Translates English, Spanish and French with Streamlit and Python Code.

Build a lightweight, offline-capable translation app for English, Spanish, and French using MarianMT models.

Features

• Local Translation: No API keys or internet required after setup

Bilingual Support:

• English ↔ Spanish
• English ↔ French
• Small Models: ~150MB per language pair
• Simple UI: Built with Streamlit for ease of use

Models & Performance

• Models Used: MarianMT from Helsinki-NLP

• opus-mt-en-es (English ↔ Spanish)

• opus-mt-en-fr (English ↔ French)

• Model Sizes: ~150MB each (total ~300MB for all pairs)

Performance:

• Best for short/medium sentences
• Moderate accuracy (comparable to mid-tier translators)
• CPU-friendly (no GPU required)

Installation

Clone the repo: git clone https://github.com/ErikElcsics/Build-Language-Translator-Local-Streamlit--Translates-English-Spanish-French-Python-Code.git cd streamlit-translator

Install dependencies:

pip install streamlit transformers torch sentencepiece

Run the app:

streamlit run LocalLanguageTranslator.py

How It Works

• Select Languages: Choose source/target from dropdowns.
• Enter Text: Type/paste text to translate.
• Click Translate: Output appears instantly.

Basic Technical Information:

• Uses Hugging Face’s transformers to load MarianMT models.
• Tokenizes input → generates translation → decodes output.

Libraries Used

• streamlit (UI)
• transformers (model loading)
• torch (PyTorch backend)
• sentencepiece (tokenization)

Notes

First run downloads models (~5-10 minutes depending on internet). Translations may be slower on CPU (2-5 seconds per sentence).

License: MIT

English to Spanish

English to French

French to English

Spanish to English