Turn a PDF into an editable Word document without handing your file to a conversion service. The catch with this conversion is that a PDF has no structure to translate: it stores glyphs at coordinates, not headings and paragraphs, so anything worth calling a converter has to infer the document back from its typography. This one groups the text into lines by position, works out the body font size by how much text is actually set in it, and promotes short lines set noticeably larger into real Word headings — so the outline pane and navigation work in the finished file. Bulleted and numbered lines become real lists, wrapped lines are rejoined into their paragraphs, and a change of size ends a paragraph rather than gluing a caption onto the prose above it. Convert as many PDFs as you like at once and take them away as a ZIP. If a file turns out to be scanned images with no text layer, it says so and points you at OCR instead of handing you an empty document. Everything runs in your browser.
| Content | Result |
|---|---|
| Text paragraphs | Faithful, reflowed as editable paragraphs |
| Headings | Real Word heading styles, from size heuristics |
| Bulleted and numbered lists | Rebuilt as Word lists |
| Multi-column pages | Correct order, columns read in sequence |
| Tables | Text in order, grid structure lost |
| Exact styling and spacing | Approximate — not recoverable from a PDF |
| Scanned pages | No text — needs OCR, and you are told so |
No. The PDF is parsed and the .docx is built in your browser. Contracts, medical records and financial statements — exactly the documents people need to edit and least want to upload — never leave your device.
No, and no converter can honestly promise that. Text is faithful and the structure — headings, paragraphs, lists — is reconstructed, but exact spacing, multi-column layouts and complex tables are approximate. A PDF stores positioned glyphs rather than a document model, so the original layout is not there to recover.
By font size relative to the rest of the document, because a PDF has no heading markup to read. The body size is whichever size carries the most text; short lines set 1.5x, 1.25x and 1.12x larger become Heading 1, 2 and 3. A long line is never promoted however large it is set, since that is a pull-quote rather than a heading.
A scanned PDF is a picture of a page with no text layer, so there is nothing to extract. You need OCR, which recognises characters from the image. The tool detects this and tells you rather than handing back an empty document.
Their text comes across in the right order, but as paragraphs rather than a Word table. PDFs do not record tables as tables — only as glyphs that happen to line up — so rebuilding the grid reliably is a substantially harder problem than reflowing prose.
It is a standard Office Open XML .docx, so Microsoft Word, Google Docs, LibreOffice Writer, Apple Pages and anything else that reads Word files.
Yes. Drop in as many files or a whole folder, and take the results away individually or as a single ZIP.
A PDF carries no document structure. There is no heading element, no paragraph, no list — only glyphs with coordinates and a font size, which is why converting one to Word is a reconstruction problem rather than a translation problem. pdf.js supplies each page's text as positioned fragments; those are grouped into visual lines by their y coordinate, sorted top to bottom (PDF user space grows upward, so the largest y is the top of the page) and left to right within a line. Structure is then inferred from typography. The body size is taken as the size carrying the most characters in the document, weighted by length rather than by line count, so three large headings cannot outvote the prose they sit above. A short line set at 1.5x the body becomes Heading 1, 1.25x Heading 2, 1.12x Heading 3; a long line is left as body text no matter how large, because a big block of text is a pull-quote rather than a heading. Lines opening with a bullet glyph or a number-and-delimiter become list items with the marker stripped. Wrapped lines are joined back into their paragraph, and a change of font size ends the paragraph, since wrapped text is set at one size and a different size means a different block. The resulting blocks are written into a real Office Open XML file by the docx library, with Word's built-in heading and list styles so the outline pane and navigation work. Both the PDF parsing and the .docx generation happen in the page; the document is never uploaded.