SnipText

Screen text extractor with OCR and spell correction for Arch Linux.

Simple workflow: Hotkey → Select area → Text in clipboard

Features

• Adaptive OCR: Automatic mode selection for speed and accuracy
• Multi-language: 100+ languages via Tesseract (English, Russian, Greek, Math symbols, etc.)
• Spell correction: Automatic text correction for better results
• Works everywhere: Wayland and X11 support
• GPU acceleration: CUDA support when available

Installation

yay -S sniptext

Important: Set up a keyboard shortcut after installation. See KEYBINDINGS.md

Usage

1. Press your keybind (default: Ctrl+Alt+T)
2. Select screen area
3. Text copied to clipboard

Test: sniptext --capture-now

Language Support

Install additional language packs:

sudo pacman -S tesseract-data-rus     # Russian
sudo pacman -S tesseract-data-ell     # Greek
sudo pacman -S tesseract-data-equ     # Math symbols
# See all: pacman -Ss tesseract-data

Update config: ~/.config/sniptext/config.yaml

ocr_language: eng+rus+equ  # English + Russian + Math

Full guide: LANGUAGES.md

Configuration

Config: ~/.config/sniptext/config.yaml (auto-created on first run)

ocr_engine: ensemble        # ensemble, tesseract, or easyocr
ocr_language: eng           # See LANGUAGES.md for codes
adaptive_ensemble: true     # Auto quality-based mode selection
enable_text_correction: true
notification_enabled: true
use_gpu: true               # CUDA acceleration

Adaptive Ensemble: Automatically uses fast mode for clear images, accurate mode for difficult ones.

Note: Optional features (spell correction, EasyOCR, ML analysis) are auto-detected when installed.

Optional Dependencies

All optional dependencies are auto-detected. Install to enable features:

yay -S python-symspellpy      # Spell correction (English)
yay -S python-scikit-learn    # Quality analysis
yay -S python-easyocr         # High-accuracy OCR (slower, GPU recommended)