GSM (GameSentenceMiner)

Turn your gaming time into language mastery.

---

🎮 See it in Action

• OCR to get get text from a game that doesn't support text hooks.
• Look up words with Yomitan in game.
• Create Anki cards with game audio + screenshot (or gif) automatically.

---

What does it do?

GSM is an application designed to automate the process of creating flashcards while you play. It sits between your game and Anki, handling audio recording, screenshots, and OCR so you don't have to interrupt your gameplay.

📝 Anki Card Enhancement

GSM automatically adds context to your Anki cards whenever you create them.

• Audio Capture: Uses Voice Activity Detection (VAD) to record and trim the specific voice line associated with the text.
• Screenshots: Captures the game state the moment the line is spoken. GIFs and Black Bar Removal are supported.
• Mine from History: Go back and create cards from previous lines you've encountered (i.e. cutscenes).
• Multi-Line Support: Capture multiple lines of dialogue into one card using the built-in Texthooker.
• AI Translation: Optional integration to provide sentence translations using your own API key.

https://github.com/user-attachments/assets/df6bc38e-d74d-423e-b270-8a82eec2394c

👁️ OCR (Text Recognition)

For games that don't have a text hook (Agent/Textractor), GSM uses a custom fork of OwOCR to read text directly from the screen.

This opens up all kinds of posssibilities for games that would otherwise be inaccessible for language learning/sentence mining. For example I've made cards with games like Metal Gear Solid 1+2, Titanfall 2, and Sekiro, all using GSM's OCR.

• Easy Setup: Managed installation means you don't need to fiddle with terminals.
• Two-Pass System: Clean, fast output similar to as if you had a hook.
• Customizable Capture Zones: Define exactly where text appears on your screen for optimal results.

https://github.com/user-attachments/assets/07240472-831a-40e6-be22-c64b880b0d66

🖥️ Overlay

GSM includes a transparent overlay for instant dictionary lookups.

• Hover over characters in-game to see definitions via Yomitan.
• Create cards without ever leaving the game window.
• Automatically Generated Furigana Display In Game.

📊 Statistics

Track your immersion habits with the stats dashboard.

• Kanji Grid: View every Kanji you've encountered and click them to see their source sentences.
• Goals: Set daily reading targets.
• Tools: Clean up and organize your mining history.

---

🚀 Getting Started

1. Download: Get the latest release.
2. Install: Watch the Installation Guide.
3. Requirements:
   • An Anki tool (Yomitan, JL, etc.)
   • A text source (Agent, Textractor, or GSM's built-in OCR)
   • A game

📚 Documentation

For full setup guides and configuration details, check the Wiki (Currently WIP).

❤️ Acknowledgements

• OwOCR & MeikiOCR for the OCR backend.
• Renji's Texthooker & Saplling.
• exSTATic for the stats design inspiration.
• Jiten, VNDB, Anilist, and IGDB for stats metadata.
• Jiten Reader which is available in GSM's overlay.
• chaiNNer for the Python integration strategy.
• Yomitan for the integrated popup dictionary overlay.
• LunaTranslator for text hooking support.
• Textractor for text hooking support.
• Agent & Agent Scripts for game text hooking support.
• Sugoi Hook for text hooking inspiration and CLI binaries.

Integrated Components

This project includes modified versions of the following libraries, I got tired of submodule hell so I've included them directly here for easier management all credits go to the original authors:

• TextFeed Page

  • GSM: https://github.com/bpwhelan/GameSentenceMiner/tree/main/texthooker
  • Original: Renji-XD/texthooker-ui

• OwOCR

  • GSM: https://github.com/bpwhelan/GameSentenceMiner/tree/main/GameSentenceMiner/owocr
  • Original: AuroraWright/owocr

• MeCab Controller

  • GSM: https://github.com/bpwhelan/GameSentenceMiner/tree/main/GameSentenceMiner/mecab
  • Original: Ajatt-Tools/mecab_controller

Star History

Sponsors