ToriOCR

Free Screen OCR for Japanese, Chinese and Korean language learners.

About

ToriOCR is a free macOS desktop tool that captures text from any screen region and recognizes selected language. Look up characters in local dictionaries, hear pronunciations, translate, and export flashcards to Anki. Watch demo

Features

• Screen region capture — works system-wide over any application
• OCR with macOS Vision or PaddleOCR — Japanese, Korean, Chinese
• Built-in translation — powered by Apple Translation (macOS 15+)
• CJK character dictionaries — KANJIDIC2, CEDICT, KEDICT with detailed entries
• Text-to-speech — TTS pronunciation for captured text
• Anki flashcard export — one-click export via AnkiConnect (preset)
• System tray app — always accessible, always-on-top results window
• Light and dark themes — follows your system appearance

Requirements

• macOS 15.0 (Sequoia) or later
• Apple Silicon (M1 or newer)

Installation

1. Download the latest .dmg from Releases
2. Open the DMG and drag ToriOCR to your Applications folder
3. Launch from Applications or Spotlight

Getting Started

1. Install ToriOCR, enable required permissions and login
2. Launch ToriOCR — it runs in your menu bar
3. Press Command + Shift + S to start a capture (change in Settings)
4. Select a screen region
5. View OCR results in the results window
6. Download languages for translation and TTS as needed
7. Click characters to see dictionary entries, translations
8. In case of errors refer to troubleshooting page

Anki Integration

ToriOCR can export flashcards directly to Anki:

1. Install Anki
2. Install the AnkiConnect plugin
3. Make sure Anki is running, select deck in Settings
4. Use the export button in the results window

Privacy

• Analytics are opt-out and collect only anonymous usage statistics
• No personal data is collected or transmitted

License

Full End User License Agreement agreement (EULA) before installation.

Acknowledgments

ToriOCR is built with Tauri & React. Dictionaries created with the help of these projects and resources:

• KANJIDIC2 — Japanese kanji dictionary (CC BY-SA 4.0)
• CEDICT — Chinese-English dictionary (CC BY-SA 4.0)
• KEDICT — Korean-English dictionary
• OAR-OCR — PaddleOCR on ONNX Runtime (Apache 2.0) © Wang Xin
• ORT — ONNX Runtime Rust bindings (MIT / Apache 2.0) © pykeio
• ONNX Runtime — (MIT) © Microsoft Corporation
• PaddleOCR Models — Pre-trained ONNX models (Apache 2.0) © PaddlePaddle Authors