pantsudango/dango-translator

★ 8,589PythonAudience · generalComplexity · 3/5LicenseSetup · moderate

Mindmap

mindmap
  root((repo))
    What it does
      Screen OCR
      Real-time translation
      Manga image mode
    OCR options
      Offline on-device
      Xingyun cloud
      PaddleOCR engine
    Translation engines
      Online AI services
      Local AI model
      Conventional engines
    Use cases
      Japanese manga
      Untranslated games
      Batch image files

mindmap root((repo)) What it does Screen OCR Real-time translation Manga image mode OCR options Offline on-device Xingyun cloud PaddleOCR engine Translation engines Online AI services Local AI model Conventional engines Use cases Japanese manga Untranslated games Batch image files

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Read untranslated Japanese manga in real time by drawing a selection box around speech bubbles to get instant translation.

USE CASE 2

Play Japanese PC games with no English patch using the screen translation overlay without switching apps.

USE CASE 3

Batch-convert raw Japanese manga image files into translated images with the original text replaced in place.

Tech stack

PythonPaddleOCRGo

Getting it running

Difficulty · moderate Time to first run · 30min

Windows only, the Python version in this repo is outdated, download the current Go-based release from the official website for new features.

LGPL license: you can use and distribute this software freely, but if you modify the library itself you must release those changes under the same license.

In plain English

Dango Translator is a Windows desktop tool that lets you point at any region of your screen, have it read the text there automatically using OCR (optical character recognition), and then immediately show you a translation. The main use case is reading raw Japanese manga or video games that have no official translation: you draw a box around the text you want translated, and the software handles the rest in real time. The tool supports both offline OCR (running entirely on your computer) and online OCR through a cloud service called Xingyun. For translation itself, you can choose from conventional translation engines, online AI translation services, or a locally running AI model. Whichever combination you pick, results appear on screen as you play or read, without switching apps. Beyond live screen translation, there is a dedicated manga mode that takes image files, detects all the text bubbles, translates them, erases the original Japanese text, and pastes the translation back in its place. This produces a translated image you can read like a normal page. The tool also includes an account system so your settings sync to the cloud and carry over when you reinstall. A note on the code in this repository: the Python version here was last updated at version 4.5.8. After that point, the project was fully rewritten in Go (Golang), and the Go version is what receives ongoing updates and new features. This repository is kept alive mainly so people can find the download links, which point to the official website and the latest release pages. If you want the newest version, follow those links rather than building from this source. The project is licensed under the GNU Lesser General Public License (LGPL) and credits PaddleOCR as the foundation for both its offline and online OCR pipelines.

Copy-paste prompts

Prompt 1

How do I set up Dango Translator to translate Japanese text in a PC game window in real time on Windows?

Prompt 2

What is the difference between offline OCR mode and Xingyun online OCR in Dango Translator, and which is more accurate?

Prompt 3

How do I use Dango Translator manga mode to translate a folder of raw Japanese manga images and export them with English text?

Prompt 4

Why does this repository use Python if the active development is now in Go, and where do I download the latest version?

Open on GitHub → Explain another repo

← pantsudango on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.