OCR Translator

Real-time screen OCR and translation tool powered by Claude Code CLI.

Features

• Glass Overlay: Position a transparent capture region over any text
• Hotkey Capture: Press F1 to capture and translate instantly
• Auto Mode: Automatic capture at configurable intervals (1-30 seconds)
• Model Selection: Toggle between Haiku (fast) and Sonnet (quality)
• Translation History: Keep track of all translations in a session

Requirements

Claude Code CLI - Required for translation

1. Install Node.js from https://nodejs.org/
2. Run: npm install -g @anthropic-ai/claude-code
3. Run: claude (to login with your account)

Installation

Option 1: Download Pre-built

1. Download OCR Translator.exe
2. Run it!

Option 2: Run from Source

pip install -r requirements.txt
python main.py

Usage

1. Position the overlay - Drag the purple-bordered region over the text you want to translate
2. Resize as needed - Drag the edges to resize the capture area
3. Press F1 - Capture and translate the text
4. Or enable Auto mode - Automatically capture at set intervals

Hotkeys

• F1 - Capture and translate
• Escape - Quit application

Controls

• Auto: ON/OFF - Toggle automatic capture mode
• Clear History - Clear translation history
• Toggle Overlay - Show/hide the capture region
• Haiku/Sonnet - Switch between fast and quality models
• Interval Slider - Set auto-capture interval (1-30 seconds)

Building from Source

pip install pyinstaller
pyinstaller --onefile --noconsole --name "OCR Translator" main.py

Or run build.bat on Windows.

Output: dist/OCR Translator.exe

Note

Each user needs their own Claude Code CLI login - no API keys are shared or exposed.

License

Open Source