Why is converting a scanned PDF to EPUB hard?
When a document is scanned, the PDF actually contains images of pages. The text is produced afterwards by OCR software and laid over the image as an invisible layer. That layer says nothing about where a paragraph begins, which line is a heading, or which word is italic — it holds only independent lines whose position on the page is known.
Ordinary converters take those lines as they are. The result is a file where every line is its own paragraph, page numbers land in the middle of sentences, and words are split in two. This tool rebuilds the text from scratch instead.
What happens during conversion?
| Problem | What is done |
|---|---|
| Watermarks and running heads | Short lines repeating across most pages, text in the margin bands and fonts differing from the body are stripped out |
| Words split at line ends | Hyphenation marks are removed and the halves are rejoined |
| Paragraph boundaries | Indentation is measured to find where a new paragraph starts; a paragraph crossing a page break stays whole |
| OCR letter errors | Corrected using the document's own vocabulary as evidence |
| Sections | Detected from the space left at section openings; the table of contents is built automatically |
| Cover and metadata | Cover image and title/author details are extracted from the PDF |
How is OCR correction kept safe?
Scanning software confuses letters: w is often read as vv. But correcting blindly damages real words — the Turkish word “kuvvet”, for instance, genuinely contains vv. This tool corrects using not an external dictionary but the document's own vocabulary: if the correct spelling already appears frequently in the same document it corrects, otherwise it leaves the word alone. Every change is listed on the result screen, and you can switch correction off entirely.
Frequently asked questions
What is the difference between a scanned PDF and a normal one?
In a scanned document every page is an image; the text is produced by OCR and placed invisibly over it. That layer carries no structural information at all — no paragraphs, headings or italics. This tool derives the structure back out of the page layout.
Are my files stored?
The uploaded PDF and the generated EPUB are kept in temporary storage and deleted automatically within 24 hours. Files are not shared with third parties.
What file size is supported?
PDF files up to 100 MB. A typical 300-page document converts in a few seconds.
Can PDFs without a text layer be converted?
Yes. If you upload such a file, we offer to run text recognition right on this page, on your own computer — the file is not sent anywhere for it. The language data is downloaded on first use, and long documents can take a few minutes.
Who is responsible for the files I upload?
Responsibility for uploaded files and for the conversion lies entirely with the user. Please upload only files you have the right to process.