How to Translate a Scanned PDF

Drag a scanned PDF into most translators and nothing happens: the file comes back untouched or with zero text. That is not a bug — a scan is a picture of a page, and translators can only read characters. A scanned page has no characters until OCR turns the image into text. So translating a scan is always two honest steps: recognise the text, then translate it.

Why translators reject scans

A translation engine works by extracting the text layer, sending those words to a model, then returning the translation. When a PDF is a scan, the extraction step finds nothing, because the letters are pixels, not text. Some services quietly return the original; others error out. The fix is identical either way — give the file a real text layer first.

Translate a scan in two steps

1. 1Run the OCR tool on your scanned PDF to add a text layer. Choose the document language if the tool offers it — it sharpens recognition for non-Latin scripts.
2. 2Open the translated PDF in the translate tool with the same approach as a normal document, select the source and target languages, and download the result.

The OCR step is what makes everything after it possible. The recognised PDF selects, copies and searches like a digital document, which is exactly the input the translation step needs.

Tips for better recognition on difficult scans

• Tell the OCR step the source language when known — the engine targets the right character set.
• A clean, flat, well-lit capture reads far better than a skewed or shadowed photo.
• After OCR, spot-check names, numbers and currency before translating them into a final document.

What the translation step actually sends

Unlike every other tool here, the translate tool sends the extracted text to a third-party translation service — that is what produces the translation. The extracted words are the only thing that leaves your device, and this warning is shown in the tool itself. For confidential scans, weight that trade-off before you start.