A Tool to Translate Scanned Manuals in Place

[fargo]

Hi all,

I have many old Japanese and Eastern European computers with no English manuals. I built an app to help me translate those manuals into English; some of you might find it useful!

The app takes scanned manual pages, OCRs the original text, translates it, and redraws the translation directly on the page — tables, diagrams, icons and logos stay exactly where they were, so the result still reads like the original manual. There's a browser-based editor to fix up any block the OCR or machine translation gets wrong (move, resize, rewrite, recolor), and output can be PNG/JPEG or per-page PDFs with a hidden searchable text layer.

It runs entirely on a Mac: OCR via the built-in Vision framework, translation via Apple's offline translator (macOS 15+); Google Translate is used as a fallback if you prefer.

Attached are two sample pages from the Fujitsu Car Marty manual, translated to English using this app (one saved as a searchable PDF).

Code and setup: https://github.com/techana/scanned-images-translator

 

[fargo]

The app now supports Windows and Linux. For OCR, it uses macOS Vision (macOS 15+) or Tesseract (Windows, Linux, or older Macs). BTW, Vision OCR gives better results, so I recommend using the Mac for this app.