Anki Video2Card

Anki Video2Card is a modern C++23 cross-platform desktop application designed to streamline the process of creating high-quality Anki cards directly from video files. It uses local morphological analysis and dictionary lookups to automatically extract and annotate Japanese text from your favorite anime, movies, or shows.

Features

• Modern C++23: Built with C++ standards for performance and type safety.
• Cross-Platform: Runs on macOS, Windows, and Linux.
• Beautiful UI: Built with SDL3 and Dear ImGui for a responsive and intuitive interface.
• Video Player: Integrated high-performance video player based on libmpv.
• Smart Extraction:
  • Snapshot: Instantly captures the current video frame as the card image.
  • Audio: Extracts audio clips in OGG/Vorbis format corresponding to the current subtitle or timestamp using FFmpeg.
  • Subtitles: Automatically extracts the current subtitle text.
• Local Text Analysis:
  • Morphological Analysis: Uses Mecab for accurate word segmentation and dictionary forms.
  • Furigana Generation: Automatically generates ruby text annotations for kanji.
  • Dictionary Lookups: Local JMDict dictionary for word definitions.
  • Pitch Accent: Automatically looks up and displays Japanese pitch accent patterns using NJAD database.
• Anki Integration: Connects directly to Anki via AnkiConnect to create cards automatically.
• Smart Fields: Automatically detects and fills fields like Sentence, Target Word, Furigana, Pitch Accent, and Definitions.

Screenshots

Shortcuts

• Space: Play/Pause
• Right Arrow: Seek forward 5s
• Left Arrow: Seek backward 5s
• M: Extract current scene (Image + Audio + Subtitle)

Getting Started

Installation

See the Installation Guide for detailed instructions on installing dependencies for your platform (macOS, Linux, or Windows).

Building

See the Building Guide for step-by-step instructions on building the application from source.

Quick Start

For experienced users:

git clone https://github.com/DaveMcMartin/anki-video2card.git
cd anki-video2card
python3 scripts/download_translation_model.py
python3 scripts/convert_pitch_accent.py
mkdir build && cd build
cmake ..
cmake --build .

Project Structure

• src/ - Main application source code
  • core/ - Core functionality
  • ui/ - User interface components (VideoSection, etc.)
  • ai/ - AI integration modules
  • api/ - API clients (AnkiConnect)
  • config/ - Configuration management
  • language/ - Language utilities
  • utils/ - Utility functions
• cmake/ - CMake build scripts and utilities
• assets/ - Application assets (icons, etc.)

Usage

1. Configuration:

   • Go to the "AnkiConfiguration" tab.
   • Set up your AnkiConnect URL (default is usually http://localhost:8765).
   • On the Card tab, select the Note Type, deck, and fields you want to fill.

2. Card Creation:

   • Drag and drop a video file into the Video Player section.
   • Navigate to the scene you want to mine.
   • Press M or click the EXTRACT button.
   • The application will capture the current frame, the audio clip for the subtitle, and the subtitle text.
   • A modal will appear allowing you to edit the extracted sentence or specify a target word.
   • Click "Process" to send the data to the AI for analysis.
   • Review the generated fields in the "Anki Card Settings" section.
   • Click "Add" to create the card in Anki.

FAQ

1. Which Note Type do you use? The "Japanese Sentence" note type from the Ankidrone Foundation deck is recommended, but you can map fields to any note type.

Dependencies

• SDL3: Cross-platform graphics and input handling.
• ImGui: Immediate-mode GUI framework with docking support.
• nlohmann/json: Modern JSON library for C++.
• cpp-httplib: Lightweight HTTP client library.
• FFmpeg: Video and audio processing library (with Vorbis codec support).
• libmpv: Media player library.
• Mecab: Japanese morphological analyzer.
• SQLite3: Database engine for dictionary and pitch accent lookups.

License

This project is licensed under the GNU General Public License v3.0 (GPLv3). See the LICENSE file for details.

Planned Features

This is a work in progress. Here are some planned features, it is not in priority order:

• Github Action on release that generates binaries for Windows, macOS, and Linux.
• Implement pitch accent generation from NJAD database
• Add support for multiple dictionary sources including local and online sources
• Improve furigana handling for edge cases
• Audio fetching from online sources (Forvo, NHK, etc)

Contributing

Contributions are welcome! Please feel free to submit issues and pull requests to improve the project.