How to convert a scanned PDF to EPUB with OCR

Last updated: 4 November 2026

To convert a scanned PDF to EPUB, you run optical character recognition (OCR) on the page images. OCR turns the pictures of text into real, reflowable text, which is then split into chapters and packaged as an EPUB. The result is a text-only book that you can resize, search and navigate. It will not look like the scan, and you should proofread it before trusting it.

What actually happens when you convert a scanned PDF to EPUB?

A scanned PDF holds a photograph of each page, not text. An EPUB needs real characters, so the converter has to read those images with OCR, rebuild paragraphs and chapters, and write the result as reflowable text.

Here is the usual pipeline:

• Page images are analysed. The software finds the lines of text and the reading order.
• Characters are recognised. Each word is guessed from its shape, using a language model for the chosen language.
• Text is rebuilt into a book. Lines are joined into paragraphs, headings are identified, and chapters become entries in the table of contents.
• The EPUB is packaged. Metadata such as title and author is added.

This is different from converting a "text PDF", where the characters already exist and nothing needs to be recognised. Our guide to converting a PDF to EPUB for free covers that simpler case.

Is your PDF really a scan?

Check first, because OCR is slow and unnecessary if the PDF already contains text. A PDF is image-only if you can't select or search any of its text.

Try this in under a minute:

• Open the PDF and press Ctrl+F (Cmd+F on a Mac). Search for a word you can see on the page.
• Try to highlight a sentence with your cursor.
• If the search finds nothing and you can only drag a box over the page like a picture, it is image-only.
• If the search works but the text is garbled, the file has a bad hidden text layer from an earlier OCR pass. Treat it as a scan.

On our converter, OCR is meant for scans with no text at all. The setting says it is slow and the result is text-only. It also says OCR is skipped automatically when it isn't needed. So switching it on for a normal text PDF should not cost you anything.

How do you convert a scanned PDF to EPUB step by step?

Upload the file, switch on "Read scanned pages (OCR)", choose the language of the book, and convert. Preparing the file and checking the settings first makes a visible difference to accuracy.

Before you upload

• Check the scan quality. OCR engines generally work best on scans of about 300 dpi. Very low-resolution scans and phone photos taken at an angle produce more errors.
• Check straightness and contrast. Skewed pages, dark gutter shadows and faded print all cause misreads.
• Know the file limit. Our converter accepts files up to 100 MB. A long book scanned in colour can exceed that, so re-save it in greyscale or at a lower image quality if needed.
• Know the language. OCR needs to be told what it is reading.

On the converter

• Drop the PDF into the upload box.
• Under Scanned pages, turn on Read scanned pages (OCR).
• Set the OCR language. The options on the page are English, Turkish, German, French, Spanish, Italian, Portuguese, Dutch and Greek.
• Under The book, fill in the title and author so they appear correctly in your library.
• Leave the mode on Auto, which picks reflowable text. Fixed layout is the other choice.
• Pick your target device: tablet or general, 6" e-ink, 7" e-ink, or generic e-ink.
• Press Convert to EPUB, then download the file.

Whether the Standard (free for text PDFs) or Pro (2 credits per page) model applies to scans isn't stated on the converter page. Check the pricing page before converting a long book. Creating an account gives you 30 credits free, according to the site.

What will the EPUB look like compared with the scan?

Expect clean, reflowable text, not a copy of the page. Pictures, decorative layouts and exact typography from the scan do not carry over, because the output is text-only.

This is the trade-off. A scan keeps the look of the page, and an EPUB keeps the words. If the illustrations matter more than the text, a scan may remain the better format. Our comparison of EPUB vs PDF explains when each format is the better choice.

How accurate is OCR, and what mistakes should you look for?

OCR accuracy depends mostly on the scan: clean, high-resolution pages of modern printed text give the best results. No general accuracy figure is established for all books, so assume some errors and proofread. Old typefaces, stains, bleed-through and tight bindings raise the error rate.

The usual mistakes are predictable:

• Look-alike characters. The letter "l" read as the digit "1", "O" as "0", "rn" as "m".
• Hyphenation. Words split at line ends may keep a stray hyphen mid-word.
• Ligatures and accents. Characters such as "fi" or accented letters can be lost, especially in a language that you haven't selected.
• Wrong language. Selecting English for a French book damages every accent and many words.
• Stray header or footer text. A book title or page number repeated inside paragraphs.
• Merged or split paragraphs. Common on pages with indented quotes or poetry.

A quick proofreading checklist

• Open the table of contents and click a few entries to confirm they land on the right chapter heading.
• Read the first and last page of two or three chapters, where headings and breaks cause most problems.
• Search for the digit "1" and for " rn " type oddities in a chapter you know well.
• Change the font size on your device and confirm that the text reflows.
• Skim a page with a table, list or quotation.

If the table of contents is missing or wrong, our guide on fixing an EPUB with no table of contents shows how to repair it.

What if the scan is poor, or you need more than text?

Improve the source file first, since a better scan beats any setting. If the page contains equations, columns or illustrations that you need, plan for manual work or keep the PDF alongside the EPUB.

• Rescan at a higher quality if you have the original. Flat pages, even lighting and 300 dpi make a difference.
• Split a huge file into parts if it exceeds 100 MB, then convert each separately.
• Crop out margins and gutter shadows in a PDF tool before you upload.
• Use the right language for the OCR step, and for mixed-language books pick the main one and expect errors in the rest.
• Don't expect equations to survive. OCR reads them poorly. Our textbook PDF to EPUB guide covers technical material in more detail.

Our page notes that uploaded files are deleted automatically. It also says desktop conversion stays local, which may matter if the scan is a private document. Only use scans of books you own, or that are in the public domain or otherwise free for you to convert.

If you have a scanned book to turn into a reflowable EPUB, you can upload it to Convert2EPUB and switch on OCR in the Scanned pages settings.

Frequently asked questions

Can you convert a scanned PDF to EPUB for free?

Our converter lists the Standard model as free for text PDFs, and an account comes with 30 free credits. Whether OCR on a scan falls under the free model isn't stated on the converter page, so check the pricing page first.

Will the EPUB keep the pictures from my scan?

No. OCR output is text-only, so illustrations and photos are not carried across. If images matter, keep the original PDF as well.

Why is OCR slow?

The software has to analyse every page image and recognise each character, which takes far more work than reading text that already exists. Longer books and higher-resolution scans take longer. That is also why OCR is skipped automatically when a PDF already contains text.

Can I read the converted EPUB on a Kindle?

Yes. Amazon's Send to Kindle accepts EPUB as it is, so you don't need to convert again. Our guide on sending an EPUB to your Kindle walks through it.

What language should I choose for OCR?

Choose the language the book is printed in. The converter offers English, Turkish, German, French, Spanish, Italian, Portuguese, Dutch and Greek. The wrong choice causes errors in nearly every word.