Game Screen Text Reader

This Python script captures game screenshots, extracts text using an LLM (Ollama), and reads the extracted content aloud using TTS (Text-to-Speech). It is designed as a playful assistant for a 5-year-old, explaining game text in a fun and engaging way.

Features

• LLM-Based Text Recognition: Extracts and interprets game text using an Ollama-based model.
• Text-to-Speech (TTS): Reads out recognized text in a natural-sounding voice.
• Screenshot Capture: Takes a screenshot of the active screen and processes the image.
• Two Modes:
  • F9 (Full Analysis): Extracts text with contextual interpretation for kids.
  • F12 (Simple Extraction): Extracts text only, without added context.
• Keep-Alive Mechanism: Ensures the LLM models remain responsive.
• Hotkey Controls:
  • F9 - Run full analysis pipeline
  • F12 - Run simple text extraction
  • ESC - Exit the program

Requirements

• Python 3.x
• Dependencies:

  pip install requests pyttsx3 keyboard pyautogui comtypes
  

Usage

1. Run the script:

   python script.py
   

2. Press F9 for full text interpretation or F12 for raw text extraction.
3. Press ESC to exit.

Configuration

• Update the endpoint variable in analyze_image_with_llm to match your Ollama server.
• Adjust the TTS settings in speak_response to customize voice and speed.

Notes

• Ensure the LLM service is running and accessible before using the script.
• The keep-alive thread helps maintain API responsiveness.
• Designed for Windows with pyttsx3 but may require adjustments for macOS/Linux.