PDF to Word
Drop a PDF here or click to choose
Choose a PDF. Password-protected files are refused.
Copy the PDF's real text layer into a DOCX or TXT file, one line per paragraph. It does not rebuild columns, tables, headers, or the page design. English OCR is optional for scans.
Read the guide: What PDF to Word actually copies
How to use
- Choose a PDF that has real text. If highlighting words works in a normal viewer, this tool can read them.
- Leave OCR off for digital PDFs. Turn it on only for pages that are pictures of paper, and only for English.
- Download DOCX or TXT. The DOCX is one paragraph per line, with a page break between PDF pages.
What you get, and what you do not
Lines are grouped from the text positions in the PDF. A heading and a body line stay separate. A two-column article is not reconstructed as columns. Tables come out as lines of text, not Word tables. Use PDF to Excel if you mainly need guessed columns.
English OCR is Tesseract in the browser. It misses small type, handwriting, and any language other than English. The DOCX marks those pages with [English OCR] so you know they were guessed. OCR needs the worker to load, and Chrome blocks that worker when you open this file directly (file://). The live site can run it.
Custom font encodings sometimes produce the wrong letters even when a viewer displays them correctly. If the preview looks like nonsense, the PDF does not expose a usable text layer. Password-protected files are refused. The limit is 80 pages so a long book does not freeze the tab.
A worked example
A 4-page digital invoice exported from accounting software usually has a real text layer. Highlighting a total in Preview or Chrome works. This tool then returns those lines as paragraphs: company name, address lines, then each item. The two-column header does not stay as two columns, and the logo is not placed. You can edit the amounts in Word. You cannot expect the DOCX to look like the invoice.
A phone scan of the same invoice is a picture. With OCR off, the page is empty. With English OCR on, you may get most of the larger type and miss the footnotes. Check the numbers before you rely on them. A contract in German, Polish, or Japanese will not be read by this OCR.
When not to use this
- You need the original layout, text boxes, or floating images. Use a desktop PDF editor.
- The important part is a table. Try PDF to Excel, and still check the columns.
- The scan is not English, or the writing is by hand.
- The PDF is longer than 80 pages. Split it and run the parts.
Questions
Why is a scanned contract empty?
A scan is a picture. There is no text to copy until OCR runs. Tick the English OCR box. On a double-clicked file, Chrome usually blocks that worker. The live site can run it.
Will Word show the same layout?
No. Headers, footers, text boxes, and floating images are not placed. You get the reading order inferred from positions.
Why are some letters wrong when the PDF looks fine?
Some PDFs draw the right shapes with a custom encoding and do not expose the real characters. If the preview is nonsense, the text layer is not usable. OCR will not fix a digital page that already has a bad text layer unless that page has no selectable text at all.