🎯 The Ultimate Translation Workbench — Context-aware AI with intuitive 2-Layer Prompt Architecture, AI Assistant, project termbase system with automatic extraction, and specialized modules.

Current Version: v1.9.50 (December 18, 2025) Framework: PyQt6 Status: Active Development

• Project Context — Complete project reference
• Architecture — System design
• Legacy Versions — Historical information
• Similar Apps — CotranslatorAI, TWAS Suite, and other translation tools
• Stargazers — A page that lists all the users who have starred this repository

python Supervertaler.py

• Hands-Free Translation: Say "next segment", "confirm", "source to target", "translate" and more
• Always-On Listening: VAD-based continuous listening - no need to press F9 twice
• Dual Recognition: OpenAI Whisper API (recommended) or local Whisper model
• Grid Toggle Button: 🎧 Voice ON/OFF button in toolbar for easy access
• Status Indicators: Status bar shows 🟢 Listening → 🔴 Recording → ⏳ Processing
• AutoHotkey Integration: Control external apps (memoQ, Trados, Word) by voice
• Custom Commands: Add your own voice commands with fuzzy matching
• Configure: Tools → Supervoice tab

• Complete Dark Theme: Full dark mode support with consistent styling across the entire application
• Compare Boxes: Translation Results panel properly displays dark backgrounds in dark mode
• Termview Visibility: All words visible in dark mode, not just terms with matches
• Access: View → Theme Editor → Select "Dark" theme

• Ctrl+K Now Opens Superlookup: All concordance searches consolidated into Superlookup - one hub for TM, Termbase, Supermemory, MT, and Web Resources
• Dual-View Toggle: Switch between Horizontal (table) and Vertical (list) layouts for TM results
• Tab Reorganization: "Project Resources" tab now before "Prompt Manager", removed redundant Concordance and Import/Export tabs
• FTS5 Full-Text Search: Concordance now uses SQLite FTS5 for 100-1000x faster search on large databases

• Language Filtering: From/To dropdowns filter TM and termbase searches by language pair
• Search Direction: Both/Source only/Target only radio buttons for precise concordance
• Yellow Highlighting: Search terms highlighted in results with compact display and tooltips

v1.9.32 - Trados SDLRPX Status Fix:

• 📦 Trados SDLRPX Status Fix - Fixed critical bug where exported SDLRPX return packages kept segments in "Draft" status instead of updating to "Translated".

v1.9.30 - Critical LLM Fix:

• 🐛 Fixed OpenAI Translation - Removed hardcoded debug path that caused "No such file or directory" errors
• 📝 Spellcheck Integration - Built-in spellcheck with 8 languages bundled (EN, NL, DE, FR, ES, PT, IT, RU)
• 📚 Optional Hunspell - Add more languages with .dic/.aff dictionary files
• 💬 Right-Click Menu - Spelling suggestions, Add to Dictionary, Ignore
• 💾 Project Settings - Spellcheck state saved per-project in .svproj files

v1.9.28 - Phrase DOCX Support & Show Invisibles:

• 📄 Phrase (Memsource) Bilingual DOCX - Full round-trip support for Phrase TMS files
• 👁️ Show Invisibles - Display spaces (·), tabs (→), NBSPs (°), line breaks (¶) in the grid
• 🎨 Smart Handling - Copy/paste, word selection, and navigation work correctly with invisibles shown

v1.9.27 - Simple Text File Import/Export:

• 📄 Simple TXT Import - Import plain text files where each line becomes a source segment
• 📤 Simple TXT Export - Export translations as matching text file with target text
• 🌐 Encoding Support - UTF-8, Latin-1, Windows-1252 with automatic detection
• 📝 Line-by-Line - Perfect for translating simple text content

v1.9.26 - Automatic Model Version Checker:

• 🔄 Auto Model Detection - Automatically checks for new LLM models from OpenAI, Anthropic, and Google
• 📅 Daily Checks - Runs once per 24 hours on startup (configurable)
• 🎨 UI Standardization - All 68 checkboxes now use consistent green (16x16px) design

v1.9.25 - Linux Compatibility:

• 🐧 Full Linux Support - Works perfectly on Ubuntu and other Linux distributions
• 📦 Complete Dependencies - One-command installation with requirements.txt

v1.9.20 - Trados Studio Package Support:

• 📦 SDLPPX Import - Import Trados Studio project packages directly
• 📤 SDLRPX Export - Create return packages for delivery back to Trados users
• 💾 Project Persistence - Save/restore SDLPPX projects across sessions

v1.9.18 - Supermemory Concordance Integration:

• 🔍 Semantic Concordance - Concordance Search (Ctrl+K) now includes Supermemory tab
• 🧠 Two-Tab Interface - TM Matches (exact) and Supermemory (meaning-based)

v1.9.17 - Supermemory Enhancements:

• 🧠 Domain Management - Categorize TMs by domain (Legal, Medical, Patents, Technical, etc.)
• 🔍 Semantic Search - Find translations by meaning using AI embeddings
• 🌐 Language Filtering - Filter by language pairs with dynamic column headers
• 🔗 Superlookup Integration - New Supermemory tab for unified lookup
• 📤 Export Options - Export to TMX or CSV format

v1.9.16 - Local LLM Support (Ollama):

• 🖥️ Offline AI - Run translation entirely on your computer with no API costs
• 🔧 Hardware Detection - Automatic model recommendations based on RAM/GPU
• 📦 Setup Wizard - One-click Ollama installation and model downloads

v1.9.15 - Bilingual Table Export/Import:

• 📋 With Tags Export - Export bilingual table with Supervertaler tags for proofreading (can be re-imported)
• 📄 Formatted Export - Export bilingual table with applied formatting for clients/archives
• 🔄 Import Changes - Re-import edited bilingual table to update translations with diff preview
• 📊 5-Column Table - Segment #, Source, Target, Status, Notes - perfect for review workflow

v1.9.14 - DOCX Export & Keyboard Navigation:

• 📤 Formatting Preservation - Export properly converts <b>, <i>, <u> tags to Word formatting
• ⌨️ Ctrl+Home/End - Navigate to first/last segment even when editing in grid cells

v1.9.13 - Document Preview & List Formatting Tags:

• 📄 Preview Tab - New Preview tab shows formatted document with headings, paragraphs, and lists
• 🔢 List Type Detection - DOCX import properly detects bullet vs numbered lists from Word XML
• 🏷️ Short List Tags - <li-o> for ordered/numbered lists (1. 2. 3.), <li-b> for bullet points (•)

v1.9.12 - Progress Indicator Status Bar:

• 📊 Words Translated - Shows X/Y words with percentage
• ✅ Confirmed Segments - Shows X/Y segments with percentage
• 🔢 Remaining Count - Segments still needing work
• 🎨 Color Coding - Red (<50%), Orange (50-80%), Green (>80%)

v1.9.11 - Navigation & Find/Replace Improvements:

• ⚡ Quick Navigation - Ctrl+Home/End to jump to first/last segment
• 🔍 Smart Pre-fill - Find/Replace dialog pre-fills selected text
• ⌨️ Ctrl+Q Shortcut - Instant term pair saving (remembers last-used termbase)

v1.9.6 - Custom File Extensions & Monolingual Export:

• 📁 Branded Extensions - .svproj (projects), .svprompt (prompts), .svntl (non-translatables)
• 🌐 Language Selection - Monolingual DOCX import now prompts for source/target languages
• 📤 Target-Only Export - Export translated content preserving original document structure
• 💾 Project Persistence - Original DOCX path saved for reliable exports
• 📊 Preview & Configure - Review extracted terms, adjust parameters (frequency, n-gram, language)
• 🎯 Visual Distinction - Project=pink, Forbidden=black, Background=priority-based blue
• ⚡ One-Click Extraction - Extract Terms button in Termbases tab (enabled when project loaded)

v1.6.0 - Complete Termbase System:

• 📚 Professional Terminology Management - SQLite-based termbase system rivaling commercial CAT tools
• 🎨 Priority-Based Highlighting - Terms highlighted in source with color intensity matching priority (1-99)
• 💡 Hover Tooltips - Mouse over highlighted terms to see translation, priority, and forbidden status
• 🖱️ Double-Click Insertion - Click any highlighted term to insert translation at cursor
• ⚫ Forbidden Term Marking - Forbidden terms highlighted in black for maximum visibility
• 🔍 Real-Time Matching - Automatic detection and display in Translation Results panel
• 🗂️ Multi-Termbase Support - Create, activate/deactivate, and manage multiple termbases
• ⌨️ Fast Term Entry - Select source → Tab → select target → Ctrl+E to add term
• ✏️ Full Management - Edit priority, forbidden flag, definition, domain in dedicated UI

v1.5.1 - Source/Target Tab Cycling:

• 🔄 Tab Key Cycling - Press Tab to jump between source and target cells
• ⌨️ Termbase Workflow - Select term in source → Tab → select translation in target
• 🔠 Ctrl+Tab - Insert actual tab character when needed

v1.5.0 - Translation Results Enhancement + Match Insertion:

• 🎯 Progressive Match Loading - All match types now accumulate (termbase + TM + MT + LLM)
• ⌨️ Match Shortcuts - Ctrl+1-9 for quick insert, Ctrl+Up/Down to navigate, Ctrl+Space to insert
• 🏷️ Tag Display Control - Optional show/hide HTML/XML tags in results
• 📊 Smart Status - Manual edits reset status requiring confirmation

v1.4.0 - Supervoice Voice Dictation + Detachable Log:

• 🎤 Supervoice Voice Dictation - AI-powered hands-free translation input
• 🌍 100+ Languages - OpenAI Whisper supports virtually any language
• ⌨️ F9 Global Hotkey - Press-to-start, press-to-stop recording anywhere
• 🎚️ 5 Model Sizes - Tiny to Large (balance speed vs accuracy)
• 🚀 Future Voice Commands - Planned parallel dictation for workflow automation
• 🪟 Detachable Log Window - Multi-monitor support with synchronized auto-scroll

Previous Features:

• 🤖 AI Assistant Enhanced Prompts - ChatGPT-quality translation prompts (v1.3.4)
• 📊 Superbench - LLM translation quality benchmarking with adaptive project sampling (v1.4.1, formerly LLM Leaderboard v1.3.3)

v1.3.1 Features - AI Assistant File Attachment Persistence:

• 📎 Persistent File Attachments - Attached files saved to disk across sessions
• 👁️ File Viewer Dialog - View attached content with markdown preview
• 🗂️ Expandable Files Panel - Collapsible UI with view/remove buttons

v1.3.0 Features - AI Assistant + 2-Layer Architecture:

• 🤖 AI Assistant with Chat Interface - Conversational prompt generation and document analysis
• 🎯 2-Layer Prompt Architecture - Simplified from 4-layer to intuitive 2-layer system
  • Layer 1: System Prompts - Infrastructure (CAT tags, formatting, core instructions)
  • Layer 2: Custom Prompts - Domain + Project + Style Guide (unified, flexible)
• ✨ Markdown Chat Formatting - Beautiful chat bubbles with bold, italic, code, and bullets
• 🧹 TagCleaner Module - Clean memoQ index tags from AutoFingers translations
• 🎨 Perfect Chat Rendering - Custom Qt delegates for professional chat UI

v1.2.4 Features - TagCleaner Module & AutoFingers Enhancement:

• ✅ TagCleaner Module - Standalone module for cleaning CAT tool tags
• ✅ AutoFingers Integration - Tag cleaning integrated with AutoFingers engine
• ✅ Status Column Improvements - Semantic icons and better visual design

v1.2.2-1.2.3 Features:

• ✅ Translation Results Panels - All match types display correctly
• ✅ Document View Formatting - Renders bold, italic, underline, list items
• ✅ Enhanced Type Column - H1-H4, Title, Sub, li, ¶ with color coding
• ✅ Tabbed Panel Interface - Translation Results | Segment Editor | Notes
• ✅ Complete Match Chaining - Termbase + TM + MT + LLM together

Core Features:

• 🎯 2-Layer Prompt Architecture - System Prompts + Custom Prompts with AI Assistant
• 🤖 AI Assistant - Conversational interface for document analysis and prompt generation
• 🧠 Context-aware AI - Leverages full document context, images, TM, and termbases
• 🤖 Multiple AI Providers - OpenAI GPT-4o/5, Claude 3.5 Sonnet, Google Gemini 2.0
• 🖥️ Local LLM (Ollama) - Run AI translation offline, no API keys needed, complete privacy
• 🌐 Machine Translation - Google Cloud Translation API integration
• 🎨 Translation Results Panel - All match types (Termbase, TM, MT, LLM) in one view
• 🔄 CAT Tool Integration - Import/export with memoQ, Trados, CafeTran
• 📊 Bilingual Review Interface - Grid, List, and Document views
• 🔍 Superlookup - System-wide search with global hotkey (Ctrl+Alt+L)
• 📝 TMX Editor - Professional translation memory editor with database support
• 🧹 AutoFingers - Automated translation pasting for memoQ with tag cleaning
• 🔧 PDF Rescue - AI-powered OCR for poorly formatted PDFs
• 🔧 Encoding Repair Tool - Detect and fix text encoding corruption (mojibake)
• 💾 Translation Memory - Fuzzy matching with TMX import/export
• 📚 Multiple Termbases - Glossary support per project

• Python: 3.8+
• PyQt6 - Modern GUI framework
• OS: Windows, macOS, Linux
• Database: SQLite (built-in)

This repository follows a lean structure optimized for efficiency:

• ✅ Only essential source code included
• ✅ Current documentation in docs/
• ✅ Historical documentation archived in docs/archive/
• ✅ Smaller repo = faster AI processing = lower costs

For comprehensive project information, see PROJECT_CONTEXT.md.

Last Updated: October 30, 2025
Latest Version: v1.0.2-Qt (October 31, 2025)

File: Supervertaler_Qt.py
Current Version: v1.0.0 Phase 5 (October 29, 2025)

Latest Features:

• 🔍 Superlookup - Search TM from anywhere (Ctrl+Alt+L)
• 🎨 Modern UI - PyQt6 with 6 built-in themes + custom theme editor
• ⚡ Better Performance - Faster, more responsive
• 🎯 Superlookup - System-wide translation memory search
• 🤖 AutoFingers - Automated translation pasting for memoQ
• 📋 memoQ Integration - Bilingual DOCX import/export
• 💾 Translation Memory - SQLite-based with FTS5 search
• 📝 TMX Editor - Professional TM editing

File: Supervertaler_tkinter.py
Current Version: v3.7.7 (October 27, 2025)

Features:

• 🤖 LLM Integration - OpenAI GPT-4/5, Anthropic Claude, Google Gemini
• 🎯 Context-aware Translation - Full document understanding
• 📚 Unified Prompt Library - System Prompts + Custom Instructions
• 🆘 PDF Rescue - AI-powered OCR for badly-formatted PDFs
• ✅ CAT Features - Segment editing, grid pagination, dual selection
• 📝 TMX Editor - Professional translation memory editor
• 🔗 CAT Tool Integration - memoQ, CafeTran, Trados Studio
• 📊 Smart Auto-export - TMX, TSV, XLIFF, Excel

Download Latest:

• Qt Edition: Supervertaler_Qt.py (Modern, recommended)
• Tkinter Edition: Supervertaler_tkinter.py (Classic, stable)

Previous Versions: See previous_versions/ folder for archived releases

Supervertaler is a professional Computer-Aided Translation (CAT) editor designed by a 30-year veteran translator for translators.

Built with PyQt6, Supervertaler offers modern UI, advanced AI integration, complete termbase system, and specialized modules for every translation challenge.

Complete Termbase System (v1.6.0):

• 📚 Professional Terminology Management - SQLite-based with FTS5 search
• 🎨 Priority-Based Highlighting - Terms highlighted with color intensity (1-99 scale)
• 💡 Hover Tooltips - See translation, priority, forbidden status on hover
• 🖱️ Double-Click Insertion - Insert translations at cursor with one click
• ⚫ Forbidden Term Marking - Black highlighting for do-not-use terms
• 🔍 Real-Time Matching - Automatic detection in Translation Results panel
• 🗂️ Multi-Termbase Support - Create, activate/deactivate multiple termbases

AI & Translation

• 🤖 Multiple AI Providers - OpenAI GPT-4o/5, Claude 3.5 Sonnet, Google Gemini 2.0
• 🎯 2-Layer Prompt Architecture - System Prompts + Custom Prompts with AI Assistant
• 🤖 AI Assistant - Conversational interface for document analysis and prompt generation
• 🧠 Context-aware Translation - Full document context, images, TM, and termbases
• 🌐 Machine Translation - Google Cloud Translation API integration
• 🎨 Translation Results Panel - All match types (Termbase, TM, MT, LLM) in one view

Professional CAT Editor:

• 📊 Bilingual Grid Interface - Source/target cells with inline editing
• 🔄 Tab Key Cycling - Jump between source and target cells
• ⌨️ Match Shortcuts - Ctrl+1-9 for quick insert, Ctrl+Up/Down to navigate
• 📝 Document View - Full document layout with formatting
• 🏷️ Tag Display Control - Optional show/hide HTML/XML tags
• 🔍 Find/Replace - Search across segments with regex support

Translation Memory:

• 💾 SQLite Backend - Fast, reliable database storage with FTS5 search
• 🔍 Fuzzy Matching - Find similar segments with match percentages
• 📝 TMX Editor - Professional TM editor handles massive 1GB+ files
• 📥 Import/Export - TMX, XLIFF, bilingual DOCX formats
• 🔄 Auto-propagation - Repeat translations automatically

Voice & Accessibility:

• 🎤 Supervoice - AI voice dictation with OpenAI Whisper (100+ languages)
• ⌨️ F9 Global Hotkey - Press-to-start, press-to-stop recording
• 🎚️ 5 Model Sizes - Tiny to Large (balance speed vs accuracy)
• 🪟 Detachable Windows - Multi-monitor support for log and panels

Specialized Modules:

• 📄 PDF Rescue - AI OCR with GPT-4 Vision for locked PDFs
• 🧹 AutoFingers - Automated translation pasting for memoQ with tag cleaning
• 📊 Superbench - LLM translation quality benchmarking with chrF++ scoring
• 🔧 Encoding Repair - Detect and fix text encoding corruption (mojibake)
• 🔍 Superlookup - System-wide TM search with global hotkey (Ctrl+Alt+L)

CAT Tool Integration:

• 📊 memoQ - Bilingual DOCX import/export with perfect alignment
• 🏢 Trados - XLIFF import/export with tag preservation
• ☕ CafeTran - Bilingual DOCX support
• 💾 Export Formats - DOCX, TSV, JSON, XLIFF, TMX, Excel, HTML, Markdown

# Clone repository
git clone https://github.com/michaelbeijer/Supervertaler.git
cd Supervertaler

# Install dependencies
pip install -r requirements.txt

# Run application
python Supervertaler.py

1. Configure API Keys: Set up OpenAI, Claude, or Gemini credentials
2. Explore System Prompts (Ctrl+P) - Browse domain-specific specialist prompts
3. Create Custom Instructions - Define your translation preferences
4. Open a Document - Import DOCX, create segments
5. Start Translating - Use System Prompts or custom instructions
6. Export Results - Session reports, TMX, auto-export to CAT tools

• Changelog: CHANGELOG.md - Complete version history
• Legacy Versions: legacy_versions/LEGACY_VERSIONS.md - Historical information
• Project Context: PROJECT_CONTEXT.md - Complete project reference
• Website: michaelbeijer.github.io/Supervertaler

• ✅ Built by a professional translator (30 years experience)
• ✅ Designed for real translation workflows, not generic AI
• ✅ Integrates with your existing CAT tools
• ✅ Context-aware for better accuracy
• ✅ Fully open source - no vendor lock-in

• ✅ Improve translator productivity (20-40% gains documented)
• ✅ Consistent quality across your translator pool
• ✅ Works with your existing CAT tool infrastructure
• ✅ Open source means you own your workflow
• ✅ Custom training and consulting available

• 🔓 Full transparency - See exactly what the AI is doing
• 🔓 No vendor lock-in - Own your translation workflow
• 🔓 Community-driven - Contribute features, report bugs
• 🔓 Sustainable - Supported through consulting and training

• Multiple providers - OpenAI, Anthropic, Google Gemini
• Multimodal support - GPT-4 Vision for figures and context
• Batch processing - Translate entire documents at once
• Context preservation - Full document analysis before translation

• 19 System Prompts - Domain specialists (Legal, Medical, Patent, Tech, etc.)
• 8 Custom Instructions - User-defined preferences
• Prompt Assistant - Generate custom prompts from document analysis
• Markdown format - Human-readable, easy to edit

• Fuzzy matching - Find similar segments
• Context display - See source alongside match
• Segment history - Learn from previous translations
• TMX export - Industry-standard format

• Auto-reports - Session reports in HTML and Markdown
• CAT tool export - Direct memoQ and CafeTran DOCX
• Format preservation - Bold, italic, formatting maintained
• Tag safety - XLIFF tags completely preserved

• ⚡ Grid pagination - 10x faster loading (50 segments/page)
• ⚡ Smart caching - Reduce API calls with TM fuzzy matching
• ⚡ Batch translation - Process 100+ segments simultaneously
• ⚡ Responsive UI - Stays responsive during large operations

• GitHub Issues: Report bugs or request features
• GitHub Discussions: Community chat and questions
• Website: supervertaler.com
• Professional Website: michaelbeijer.co.uk

• Enhance personal productivity with AI
• Maintain consistent terminology
• Work faster without sacrificing quality
• Leverage domain-specific prompts

• Train all translators with same prompts
• Maintain company-wide consistency
• Increase productivity across the team
• Reduce review/QA time
• Custom LSP consulting available

• Learn professional translation workflows
• Understand CAT tool integration
• Practice with real-world tools
• Open source to study and modify

• No data collection - Your translations stay on your computer
• Local processing - Translations processed locally by default
• API keys encrypted - Credentials stored securely
• Open source - Full audit trail, no hidden code
• GDPR compliant - User data never leaves your system

MIT License - Fully open source and free

This software is provided as-is for both personal and commercial use.

Supervertaler is maintained by Michael Beijer, a professional translator with 30 years of experience in technical and patent translation. The project represents a personal passion for building tools that make translators' lives easier.

• 🌐 Website: michaelbeijer.co.uk
• 💼 Professional: ProZ Profile
• 🔗 LinkedIn: linkedin.com/in/michaelbeijer

Last Updated: November 16, 2025
Current Version: v1.6.6

• Enhanced Prompt Assistant with auto-refinement
• Glossary management UI improvements
• Advanced TM features (penalty weights, leverage scoring)
• Integration marketplace (partner CAT tools)
• Professional cloud hosting option (optional)

We're looking for:

• 🐛 Bug reports and feature requests
• 💡 Prompt contributions (System Prompts, Custom Instructions)
• 📖 Documentation improvements
• 🌍 Translations and localization
• 🤝 Code contributions

Check out:

1. README.md (this file) - Overview
2. CHANGELOG.md - Complete version history
3. PROJECT_CONTEXT.md - Complete project reference
4. GitHub Discussions - Ask questions & share ideas
5. Website Documentation - Guides and tutorials
6. GitHub Issues - Bug reports & feature requests

We welcome contributions and feedback from the community!

Have an idea for a new module or feature? We'd love to hear from you!

• 💬 Start a Discussion - Share ideas, ask questions, discuss features
  • Perfect for brainstorming new modules
  • Exploring "what if" scenarios
  • Getting community feedback
  • Discussing implementation approaches

Found a problem? Help us improve!

• 🐛 Report a Bug - Submit detailed bug reports
  • Include steps to reproduce
  • Specify your environment (OS, Python version)
  • Attach screenshots if relevant

1. 💭 Idea → Start in Discussions
2. ✅ Approved → Converted to Issue for tracking
3. 🚀 Implemented → Linked to commits and released

Last Updated: October 31, 2025
Version: v1.1.0 (Qt Edition)
Status: Active Development
License: MIT (Open Source)
Security Status: Current - Security patches applied

🎯 Supervertaler: Empowering professional translators with intelligent, context-aware AI tools. Built by translators, for translators.