Here is the number-one reason PDF translations fail: the PDF is a scan. A scanned PDF contains no selectable text — just pictures of pages. Translation tools read text, so feeding them a scan gives you nothing, an error, or a file that comes back unchanged. The fix is a two-step workflow: OCR first, translate second. This guide walks through exactly how to do it free, how to tell whether your file even needs OCR, and how to catch the errors that slip through when two automated steps are stacked on top of each other.
Step 0: Check whether your PDF is actually scanned
Not every PDF that looks like a scan is one, and not every clean-looking PDF has real text. Run this 10-second test before you do anything else:
- Open the PDF and try to select a sentence with your mouse.
- If you can highlight individual words and copy them into another app, it is a real text PDF. Translate it directly at PurePDF's translate tool — no OCR step needed.
- If you cannot select anything, or the whole page selects as a single image, it is scanned. Continue below.
- Mixed files exist too: some PDFs have real text on some pages and scanned images on others (common with reports that mix typed pages with photographed appendices). If only part of the document selects, treat the scanned pages separately — OCR those pages, then recombine before translating, or translate the two parts and merge the results.
Skipping this check is the most common wasted step. OCR-ing a text PDF does nothing useful, and translating a scanned PDF without OCR does nothing at all.
Step 1: OCR the scan (free)
OCR (optical character recognition) reads the page images and reconstructs a text layer. That text layer is what the translator will actually work on, so OCR quality sets a ceiling on translation quality: every character the OCR misreads is a character the translator will confidently translate wrong.
Free OCR options, honestly compared:
- PurePDF's OCR tool — upload the scanned PDF at PurePDF's OCR tool and get back a searchable PDF with the text layer embedded. The output stays a PDF with its page structure intact, which matters because the next step (translation) expects a PDF. No signup is needed for the core tools.
- Google Drive — upload the scan to Drive, right-click, then Open with Google Docs. Drive OCRs the file into an editable document. It is fully free, but formatting is usually destroyed in the process, and it struggles with longer documents. It is a good fallback when you want editable text rather than a translated PDF.
- Your phone's scanner app — most scanner apps now include OCR, but they typically export one page at a time or a plain-text dump, which makes the translate step fiddly for a multi-page document. Use this route for a single page, not a 40-page report.
For translation purposes, the PurePDF route is usually the smoothest: OCR to a searchable PDF, then translate that PDF in the next step, keeping one file through the whole workflow.
OCR quality tips that actually move the needle:
- Resolution matters. Scans at 300 DPI translate far better than phone photos taken at an angle in poor light. If you are re-scanning, use the highest resolution available and scan flat, not photographed at an angle.
- Straighten and brighten first. Crooked pages, shadows, and dark margins cause character errors that then get translated — garbage in, garbage out. Most scanner apps have an auto-straighten and contrast pass; run it before OCR, not after.
- Know the handwriting limit. Handwriting OCR is unreliable in every free tool. If the document is handwritten, type out the text manually before translating — no free OCR pass will rescue it, and a bad OCR pass will silently corrupt names and numbers.
- Check the source language setting. If your OCR tool lets you pick the document language, pick the language the scan is actually written in, not the language you want to translate into. OCR-ing Spanish text with an English model produces worse text than OCR-ing it as Spanish.
Step 2: Translate the OCR'd PDF
Once the PDF has a real text layer, translate it like any normal PDF:
- Go to PurePDF's translate tool.
- Upload the OCR'd file — no signup needed for the free tier.
- Pick your language pair and translate.
- Download the translated PDF and keep the OCR'd original alongside it, so you can compare the two page by page during the checks in Step 3.
The free tier covers short documents (the first pages free; page limits apply — check the tool page for the current free allowance). Longer scans need a paid plan: Translate Lite Pro is $5.99/month for 30 pages a day, with larger Plus and Pro tiers above it. If you only have one long document, translate it in the month you need it and cancel after — there is no reason to pay for a year of a tool you will use once.
A note on order, because it trips people up: OCR first in the source language, then translate. If your scan is in Spanish and you need English, OCR it as Spanish, then translate Spanish to English. Never try to translate first — there is nothing to translate yet — and never OCR the already-translated output, which some workflows accidentally do when they re-scan a printed translation.
Step 3: Spot-check the translation (don't skip this)
OCR plus translation is two automated steps stacked, so errors compound. A character misread by OCR becomes a wrong word in translation, and neither step flags it. Always check these before you send the document anywhere:
- Numbers, dates, and names. OCR misreads 0/O, 1/l, and 5/S constantly, and the translator then confidently translates the wrong character. Verify every figure, date, and proper name against the original scan.
- Tables. OCR often jumbles cell order, especially in multi-column tables. Verify totals and row alignment against the original — a total that moved one row over is the classic silent error.
- The first and last pages. Errors cluster at document boundaries, where covers, headers, and footers confuse both OCR and translation.
- One full paragraph, read properly. Pick a dense paragraph in the middle and read the translation against the original. If that paragraph is clean, the rest is probably usable; if it is garbled, re-OCR at higher quality rather than trying to fix the translation by hand.
For anything official — a contract, a certificate, a medical record — treat the OCR-plus-translation output as a reading aid, not a certified document. If the translation will be relied on by someone else, it needs a human translator who works from the original, not from an OCR pass.
What doesn't work (save yourself the trouble)
- Uploading the raw scan to Google Translate's document mode. It will either reject the file or return the images untranslated. This is the single most-reported failure for scanned PDFs, and it is not a bug — the tool has no text to work with.
- Screenshotting pages into an image translator. Works for a paragraph, not a 40-page document. The per-page manual loop does not scale, and page order gets lost.
- "Translate" buttons in PDF readers. They extract whatever text layer exists — on a scan, that is nothing, so they return nothing or translate only the file's metadata.
- Translating the OCR text dump and pasting it back. You lose every table, heading, and page break. Keep the file as a PDF through both steps.
FAQ
Can I translate a scanned PDF directly without OCR? No. Translation requires machine-readable text, and a scan contains only images. OCR is a mandatory first step, not an optional polish pass.
Is OCR + translation really free? Yes for short documents — PurePDF's OCR and translate free tiers both work without signup, with page limits on the free tier. Google Drive's OCR-to-Docs route is fully free but lower fidelity, since it does not preserve the PDF layout.
My scan is in Spanish and I need English — does order matter? Yes: OCR first in the source language (Spanish), then translate Spanish to English. Translating first is impossible (there is no text yet), and OCR-ing after translation just re-reads the translated pages.
Why is my translated scan full of errors when the original looks clear? Usually one of three things: the scan resolution was too low, the pages were crooked or shadowed, or the document mixes printed text with handwriting or stamps. Re-scan flat at higher resolution and re-run OCR before blaming the translator — most "translation" errors on scanned PDFs are actually OCR errors wearing a translation's clothes.
Can I translate a scanned PDF on my phone? Yes, using the same two steps in a mobile browser: OCR at PurePDF's OCR tool, then translate at PurePDF's translate tool. For a single page, a phone scanner app with built-in OCR is a fine shortcut; for a multi-page document, the browser workflow keeps page order intact.
Ready to translate pdf?
Use our free online tool to process your files securely in high quality. No sign-up required.
Open translate pdf Tool