OmniDocWhy Scanned PDFs Can't Be Translated and How to Fix It
Scanned PDFs have no text to translate because each page is an image. Learn how to confirm the problem and run OCR in OmniDoc first.
When you can't translate scanned PDFs, the translation tool usually reports that there is no text, or it returns a document with empty sections. The file itself looks fine when you open it. The pages show words, tables and signatures, but the words are part of a picture rather than characters a computer can read. Translation depends on that text, so the file has to be converted first.
This guide is for anyone who receives a scan of a contract, a form, a report or a textbook chapter and needs it in another language. It explains how to confirm that the problem is the scan, why it happens, and how to run OCR (optical character recognition) in OmniDoc so the translation can work on the words.
By the end, you will know how to check a PDF for a text layer, how to make a searchable copy, and how to decide whether the result is good enough to translate.
Why can't you translate scanned PDFs? The short answer
A scanned PDF stores each page as an image, so there are no characters for a translator to read until OCR turns that image into searchable text.
| Your situation | What to do | Why |
|---|---|---|
| You cannot select any text on the pages | Run OCR on a copy in the OmniDoc desktop client, then translate the searchable file | OCR creates a text layer the translator can read |
| The file is a phone photo of a page | Run OCR, then compare the recognized words with the page | Angles, shadows and blur lower recognition quality |
| The upload works but some lines are missing | Run OCR first and compare the output with the original | Some regions were not read, so they carry over as gaps |
| The PDF was exported from a digital document | Upload it directly for translation | Its text is already stored as characters |
| You need an editable version later | Work from the OCR copy and keep the original scan | The source stays unchanged and you get a text version |
Figure out which problem you have
Several symptoms look similar, but the causes lead to different fixes. Use these checks before you change anything.
| Symptom | Likely cause | How to confirm |
|---|---|---|
| The tool says no text was found | The pages are images only | Select a line in a PDF viewer. If nothing highlights, the file has no text layer |
| Only some pages can be selected | A mixed file, with scanned and digital pages | Check each page by selecting a line; scanned pages will not highlight |
| Searching for a visible word returns nothing | No text layer, or the text is in an image | Search for a word you can clearly see on the page |
| Translated output has blank areas | Some scanned regions were not read | Compare the output page with the original, line by line |
| Words come out garbled | Low-resolution, tilted or faint scan | Zoom in on the original; blur or tilt points to the cause |
| Table cells or columns are misread | Small print or dense layouts | Check the OCR copy against the original table |
Fix it step by step
OmniDoc accepts scanned PDFs for translation as they are, so start with a test upload. If the output is complete and readable, you can stop there. If text is missing or poor, follow these steps.
- Confirm the file is scanned. Open it in any PDF viewer and try selecting a line of text. If nothing selects, treat the file as image-only.
- Ask for a better source if one exists. If the sender has a digital original, request that file. It avoids OCR altogether. If you must work from a scan, a flat, evenly lit, sharp copy usually reads better than a photo taken at an angle.
- Install the desktop client. Download it from download.omnidoc.top for Windows, macOS or Linux, and open the scanned PDF in the PDF editor.
- Run OCR and save a new file. Apply OCR to the scanned PDF and save the result as a separate file, such as contract_ocr.pdf. Keep the original scan untouched so you can go back to it.
- Check that the text layer works. Select a line of text in the new file and search for a word you can see on the page. If both succeed, the words are now searchable.
- Compare the OCR text with the original. Focus on numbers, names, dates, codes and table cells. If a word is wrong, use the PDF editor to edit the paragraph text in place.
- Translate the searchable PDF. Upload the corrected file on omnidoc.top, choose bilingual or translation-only output, and download the result.
- Review the translation before you rely on it. Check key terms, figures, dates and any signed clauses against the source. Machine translation and OCR can both make mistakes, so important content needs a human check.
Prevent it next time
Most of the trouble comes from receiving scans that were never meant to be read by software. A few habits make the problem easier to catch early.
- Request digital originals. When you ask someone for a document, ask for the exported file rather than a scan. Keep the scan as a fallback only.
- Test the file on arrival. Select a line of text as soon as a PDF comes in. You will know within a minute whether OCR is needed.
- Scan with care. Use flat placement, even light and a sharp focus, and avoid cutting off text at the page edges.
- Name your copies. Save OCR results with a clear suffix, so the original scan and the searchable version are never confused.
- Spot-check dense pages. Tables, footnotes and small print are where OCR errors cluster, so check those pages first.
🎯 Which approach fits you?
Your situation changes the path. If you need a translated copy only once, a test upload followed by OCR for the failed pages is usually enough. If you work from scans every week, run OCR on each incoming file, keep the originals in a separate folder, and review the same kinds of pages each time.
❓ FAQ
1. Can I translate a scanned PDF without running OCR?
Sometimes. OmniDoc accepts scanned PDFs as they are, so run a test upload first. If the output is complete and readable, OCR isn't needed. If text is missing, run OCR on a copy and upload that instead.
2. Is a photo of a document the same as a scanned PDF?
For translation, yes. Both are pictures of text, so neither has characters until OCR runs. Photos taken at an angle or in shadow usually need more checking after recognition than flat scans do.
3. Why does the OCR text still contain mistakes?
Recognition depends on how clear the original page is. Small print, faint ink, skewed lines and unusual fonts cause misreads. Correct the important words before you translate, and check the translated lines that contain them.
4. Can I edit the text after running OCR?
Yes. The PDF editor lets you edit paragraph text in place, so you can fix misread words in the searchable copy. Keep the original scan as your reference.
📝 Summary
When you can't translate scanned PDFs, the cause is almost always that the pages are images with no text layer. Confirm the problem with a selection test, run OCR on a copy in OmniDoc if the test upload fails, and check the recognized text before you translate. Keep the original scan, and review important content in the final translation.
