https://github.com/danily07/translumo
Advanced real-time screen translator for games, hardcoded subtitles in videos, static text and etc.
https://github.com/danily07/translumo
autotranslate easyocr game-translation mlnet ocr translation
Last synced: about 1 year ago
JSON representation
Advanced real-time screen translator for games, hardcoded subtitles in videos, static text and etc.
- Host: GitHub
- URL: https://github.com/danily07/translumo
- Owner: Danily07
- License: apache-2.0
- Created: 2023-06-16T12:18:22.000Z (about 3 years ago)
- Default Branch: master
- Last Pushed: 2024-07-25T07:35:45.000Z (almost 2 years ago)
- Last Synced: 2025-04-03T10:06:58.658Z (about 1 year ago)
- Topics: autotranslate, easyocr, game-translation, mlnet, ocr, translation
- Language: C#
- Homepage:
- Size: 99.9 MB
- Stars: 3,694
- Watchers: 47
- Forks: 191
- Open Issues: 129
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
[](https://opensource.org/licenses/Apache-2.0)
[]()
EN | RU
Advanced screen translator. Translumo is able to detect and translate appearing in the selected area text in real-time (e.g. subtitles).
Main features
- High text recognition precision
- Simple interface
- Low latency
-
Integrated modern OCR engines: Tesseract 5.2, WindowsOCR, EasyOCR -
Available translators: Google Translate, Yandex translate, Naver Papago, DeepL -
Available recognition languages: English, Russian, Japanese, Chinese (simplified), Korean -
Available translation languages: English, Russian, Japanese, Chinese (simplified), Korean, French, Spanish, German, Portuguese, Italian, Vietnamese, Thai, Turkish, Arabic
Translumo allows to combine the usage of several OCR engines simultaneously. It uses machine learning training model for scoring each recognized result by OCR and chooses the best one.
The main idea was to make tool, that does not require manual adjustments for each case and convenient for everyday use.
There are several implemented optimizations to reduce impact on system performance and minimize latency between the moment a text appears and actual translation.
System requirements
- Windows 10 build 19041 (20H1) / Windows 11
- DirectX11
- 8 GB RAM (for mode with EasyOCR)
- 5 GB free storage space (for mode with EasyOCR)
- Nvidia GPU with CUDA SDK 11.8 support (GTX 750, 8xxM, 9xx series or later) (for mode with EasyOCR)
How to use
- Open the Settings
- Select Languages->Source language and Languages->Translation language
- Select Text recognition->Engines (please check Usage tips for recommendation modes)
- Select capture area
- Run translation
Usage tips
Generally, I recommend always keep Windows OCR turned on. This is the most effective OCR for the primary text detection with less impact on performance.
Recommended combinations of OCR engines
-
Tesseract-Windows OCR-EasyOCR - advanced mode with the highest precision -
Tesseract-Windows OCR - noticeably less impact on system performance. It will be enough for cases when text has simple solid background and font is quite common -
Windows OCR-EasyOCR - for very specific complex cases it makes sense to disable Tesseract and avoid unnecessary text noises
Select minimum capture area
It reduces chances of getting into the area random letters from background. Also the larger frame will take longer to process.
Use proxy list to avoid blocking by translation services
Some translators sometimes block client for a large number of requests. You can configure personal/shared IPv4 proxies (1-2 should be enough) on Languages->Proxy tab. The application will alternately use proxies for requests to reduce number from one IP address.
Use Borderless/Windowed modes in games (not Fullscreen)
It is necessary to display the translation window overlay correctly.
If the game doesn't have such mode, you can use external tools to make it borderless (e.g. Borderless Gaming)
Install the application on SSD
To reduce cold launch time with enabled EasyOCR engine (loading large EasyOCR model into RAM).
FAQ
I got error "Failed to capture screen" or nothing happens after translation starts
Make sure that target window with text is active. Also try to restart Translumo or reopen target window.
I got error "Text translation is failed" after successful using the translation
There is a high probability that translation service temporarily blocked requests from your IP. You can change translator or configure proxy list.
Can't enable Windows OCR
Make sure that the application is runned as an Administrator. Translumo each time tries check installed Windows language pack via PowerShell.
I set borderless/windowed mode, but a translation window is still displayed under a game
When game is running and focused use the hotkey (ALT+T by default) to hide and then show again translation window
Package downloading for EasyOCR failed
Try to re-install it under VPN
Hotkeys don't work
Other applications may intercept specific hotkeys handling
I got error "Text detection is failed (TesseractOCREngine)"
Make sure that the application path contains only latin letters.
Build
Visual Studio 2022 and .NET 7 SDK are required
- Clone repository (master branch always equals last release version):
- Restore packages and build a solution. binaries_extract.bat will be executed during building, which will automatically download models and python binaries to target output directory.
```bash
git clone https://github.com/Danily07/Translumo.git
```