PrivatePDF

Convert a PDF to Word

Most PDF-to-Word converters give you a document made of hundreds of floating text boxes. It looks like the original until you change a word, and then it falls apart — which is a strange result for a conversion whose whole purpose was to let you edit the thing. This one rebuilds the text instead: paragraphs, headings and reading order, as real Word paragraphs that re-flow when you type. It runs in your browser, so the document is never uploaded.

or drop a file anywhere on this panel

opened locally · 0 bytes sent

How it works

  1. 01Open the PDF.
  2. 02Read the panel — it tells you how many pages carry a text layer, and how columns will be handled.
  3. 03Choose whether each PDF page should start a new page in Word.
  4. 04Download the .docx and open it in Word, LibreOffice or Google Docs.

It recovers the text, not the layout

This is the honest description of what you get, and it is a deliberate trade rather than a shortcoming we intend to grow out of. Paragraphs, heading levels and reading order come across, and they behave like text somebody typed: editable, re-flowable, ready to take your own styles. Exact positions, text boxes and decorative framing do not. If what you actually need is a faithful picture of the page, convert it to PowerPoint or to images instead — those keep the appearance exactly, at the cost of the text being uneditable.

Where a line break came from matters

A PDF has no paragraphs in it. It has glyphs at coordinates, and the paragraph the author typed was dissolved into them when the file was made. So the conversion has to work out which lines belong together, from the spacing between them, where each line starts, and whether the previous one stopped short of the margin. Lines inside one paragraph are rejoined into prose, because the break was the page's and not the author's — keeping it would give you a document that re-wraps into ribbons the first time you change a margin.

Columns are read down, not across

A two-column page is where naive conversion fails worst, and fails invisibly: both columns sit on the same lines, so reading the page as a whole runs the two together into text that is fluent and wrong. This looks for the vertical white space between columns in the page geometry, splits the page there, and reads each column through in turn — so a newsletter or an academic paper comes out in the order you would read it. An unusual layout can still defeat that, so it is worth a glance. What this does get wrong elsewhere — a paragraph split in two, two short paragraphs joined — is visible immediately and quick to fix.

A scan converts to nothing, and says so

A scanned page is a photograph. There is no text in it to recover, so there is nothing to convert until it has been through OCR. The panel detects this from the page contents rather than guessing, and tells you up front; if you convert anyway, the Word document says why it is empty instead of just being blank.

Questions

Will the Word file look exactly like the PDF?
No, and that is deliberate. You get the text with its paragraphs and headings intact and editable, not a replica of the page. Converters that reproduce the appearance do it by absolutely positioning every line, which looks right until you edit a word.
Are tables and images included?
Not yet — this converts the text. For a table, use PDF to Excel, which reads ruled tables properly rather than approximating them. For images, the extract-images and PDF-to-image tools get them at full quality.
Can I convert a Word file back into a PDF here?
No, and we would rather say so than do it badly. Laying out a Word document needs a full layout engine and the original fonts; without them the output is subtly wrong in ways you would not spot until it was printed or sent.
Is the document uploaded?
No. Your browser reads the text and writes the .docx itself. You can disconnect from the network first and the conversion still works.