Blab Translation
← All guides
Documents2 minutes

Scanned PDFs and OCR: what to expect

A scanned PDF is a stack of pictures of pages: there is no text in it for a translator to read. OCR — optical character recognition — rebuilds that text first, so the translation can be laid back onto the page.

How a scan is recognised

When you choose a PDF, the browser checks how many of its pages carry a text layer and shows “Needs OCR” if it finds pages without one. That badge is an estimate made before anything is uploaded.

The decision that counts is made after upload, once, for the document as a whole. Its opening pages are sampled: if only a small share of them have text, the document is treated as a scan and every page is recognised. Otherwise it is translated from its own text layer and no page is recognised.

Step by step

  1. 01
    Import it like any PDF
    Drop the file into Documents or press “Choose a file”. There is no OCR switch to turn on.
  2. 02
    Read the badge
    “Needs OCR” means the browser found pages without text; “Ready to translate” means it did not. The “Text layer rebuilt (OCR)” row under “Before importing” is a notice, not a setting.
  3. 03
    Keep Translate on
    Recognition only runs for translation jobs. A scan imported with “Translate” switched off has no text to show, and the import fails.
  4. 04
    Start and follow it
    Press “Start translating”. Recognition comes before translation and adds time; the task in “Tasks and delivery” shows how far it has got.

Limits and charges

OCR is charged per recognised page, separately from translation, and only when the job succeeds. Because a document judged to be a scan is recognised from the first page to the last, its typed pages are counted too. The current rate is on the Pricing page.

Only PDFs are recognised. Word, EPUB, TXT, Markdown and MOBI files are read from their own text, and image files such as JPG or PNG are not accepted: put scanned pages into a PDF first.

A very faint or badly skewed scan may not be readable. The job then stops with a failure report, and the credits it reserved are returned.

What comes next

Once recognised, a scan behaves like any other document: it opens at the reading desk, keeps its layout in the translated copy, and can carry highlights, notes and terms.

Questions people ask

Can I turn OCR on or off?

No. Whether a PDF is recognised is decided from the document itself after upload; there is no switch.

Is OCR charged if the translation fails?

No. OCR is charged only when the job succeeds, and a failed job returns the credits it reserved.

Can I upload photos of pages?

Not directly. Documents accepts PDF, DOCX, EPUB, TXT, MD and MOBI files, so combine the photos into a PDF first.

Read it, then go do it — the button below lands on the screen it describes.
See import settings →