Spaces:
Running
Running
| title: DocumentTranslator | |
| emoji: 🐠 | |
| colorFrom: pink | |
| colorTo: red | |
| sdk: gradio | |
| sdk_version: 6.2.0 | |
| app_file: app.py | |
| pinned: false | |
| license: agpl-3.0 | |
| short_description: Translate Word .doxc, and .pptx, preserving layout | |
| # Document Translator | |
| Translate Word documents while preserving formatting, footnotes, and styling. | |
| ## Features | |
| - **Multiple Backends**: NLLB, Madlad-400, Opus-MT, WMT21 | |
| - **Format Preservation**: Maintains e.g. bold, italic, fonts, spacing | |
| - **Footnote Support**: Handles documents with citations | |
| - **Fast Processing**: Optimized with CTranslate2, works reasonably well on CPU | |
| ## Usage | |
| 1. Upload a .docx file | |
| 2. Select source and target languages | |
| 3. Choose translation settings | |
| 4. Click "Translate Document" | |
| 5. Download the translated file | |
| ## Configuration | |
| For optional (experimental) LLM modes, clone this Space and try with setting environment variables in Space settings: | |
| - `OPENAI_API_KEY` - For OpenAI GPT models | |
| - `ANTHROPIC_API_KEY` - For Claude models | |
| ## Limitations for now | |
| - CPU-based processing (no GPU on free tier) | |
| - NLLB 600M recommended for RAM constraints | |
| - Large documents may timeout |