🎮 Game Text-to-VoiceOver AI

Modular, privacy-first pipeline for capturing, extracting, and narrating in-game note text—designed for immersive accessibility and blind playthroughs.

🎯 Purpose

This project enables blind or visually impaired players to experience narrative-driven games by extracting in-game note text and converting it to voice. Built for modular clarity, forensic traceability, and full local control—no cloud dependencies, no hidden steps.

🧩 Features

• 🖼️ Visual Calibration: Overlay OCR zones directly on the active game window with -calibrate
• 📐 Region Isolation: Crops only the narrative zones, excluding UI clutter
• 🧠 OCR Extraction: Uses Tesseract to extract clean, readable text from game notes
• 🗣️ Voice Synthesis: Narrates extracted text using pyttsx3 with configurable rate, volume, and voice
• 🗂️ Modular Design: Each step—capture, OCR, voice—is independently testable and traceable
• 🔒 Privacy-First: All processing is local; no cloud APIs or external data sharing
• ⚙️ Unified Configuration: All settings (resolution, regions, voice, transcription) live in src/config.py
• 🧪 Test Harnesses: Lightweight scripts for OCR and voice validation
• 🧾 Daily Logging: Optional transcription logging to timestamped .txt files

📦 Installation

Requirements:

• Python 3.10+
• Tesseract OCR installed and added to PATH
• pip packages:
  • pyttsx3
  • pillow
  • pyautogui
  • pygetwindow
  • pytesseract

Optional (Windows only):

• PowerShell for timed test execution
• Tesseract installer: https://github.com/tesseract-ocr/tesseract

⚙️ Configuration

All settings are stored in src/config.py:

Resolution:

• width: 1920
• height: 1080

OCR Regions:

• Title:
  • upper_left: x=1035, y=110
  • lower_right: x=1700, y=168
• Main:
  • upper_left: x=1035, y=170
  • lower_right: x=1700, y=900

Voice Settings:

• save_voice_to_file: False
• rate: 180
• volume: 1.0
• voice_id: None

Transcription Settings:

• save_transcription_to_file: False

You can adjust coordinates, voice settings, and logging behavior without touching any core logic.

🚀 Usage

Run narration pipeline:

• python game_narrate.py

Run calibration overlay:

• python game_narrate.py -calibrate

This will draw red and blue boxes over the "Main" and "Title" OCR zones and save a screenshot to assets/calibration/zone_calibration.png.

🧪 Testing

OCR Extraction:

• Start-Sleep -Seconds 5; py -3.10 tests/ocr_test.py

Voice Synthesis:

• Start-Sleep -Seconds 5; py -3.10 tests/voice_test.py

📁 Project Structure

Game-Text-To-VoiceOver-AI/
├── src/
│   ├── capture.py         → Active window capture
│   ├── ocr.py             → Region crop + OCR
│   ├── voice.py           → Voice synthesis
│   ├── utils.py           → Logging + preprocessing
│   ├── config.py          → Unified settings and regions
├── tests/
│   ├── ocr_test.py        → OCR pipeline test
│   └── voice_test.py      → Voice synthesis test
├── assets/
│   ├── scripts/           → Daily transcription logs
│   ├── VOs/               → Saved voice files
│   └── calibration/       → OCR zone overlays
├── docs/
│   ├── architecture.md
│   ├── modules.md
│   ├── setup.md
│   ├── privacy.md
│   ├── roadmap.md
│   └── changelog.md
├── .gitignore
└── README.md

📚 Documentation

• Architecture → docs/architecture.md
• Modules → docs/modules.md
• Setup Guide → docs/setup.md
• Privacy Principles → docs/privacy.md
• Roadmap → docs/roadmap.md
• Changelog → docs/changelog.md

🛡️ Privacy & Philosophy

This project is built for forensic-grade accessibility:

• No cloud APIs
• No telemetry
• No hidden dependencies
• Every step is modular, inspectable, and overrideable

Perfect for blind playthroughs, immersive narration, and traceable pipelines.