Product
Convert a PDF into an editable Word document
Upload a PDF and download a real .docx you can edit. Text-based PDFs are read directly; scanned pages are recognised with OCR. Page order, Arabic text and right-to-left paragraphs are preserved.
How it works
Two kinds of PDF, one converter
A PDF exported from software contains real text: the characters are in the file and can be read directly, which is both fast and exact. A scanned PDF contains no text at all — each page is a picture of paper, and the words have to be recognised from the pixels.
This converter inspects each page and picks the right route on its own. A document that mixes the two, such as a printed contract with a scanned signature page, is handled page by page.
What ends up in the Word file
Paragraphs are rebuilt from the lines on the page, so a sentence that wrapped across three lines becomes one editable paragraph rather than three. Headings are detected from type size, bulleted and numbered lists keep their list formatting, and rows that share column positions are written as a real Word table.
Page order is preserved and each source page starts a new page in the document, so the result lines up with the original when you read them side by side.
Arabic and right-to-left text
Arabic paragraphs are marked as bidirectional and aligned to the right, which is what makes them read correctly in Word instead of appearing reversed or mixed up when a line contains both Arabic and Latin characters.
Direction is decided per paragraph from the characters it actually contains, so a mostly-English document with an Arabic address block gets both right.
Supported formats
- PDF with a text layer — exported from Word, an accounting package or a browser
- Scanned PDF — pages that are images of paper, read with text recognition
- Multi-page documents, converted in page order
- Latin, Arabic and other Unicode scripts
What may not carry over
- Complex page layouts — multiple columns, sidebars, text flowing around images — are converted to a single reading order, not reproduced visually.
- Handwriting is not recognised reliably and often produces nothing usable.
- Decorative elements, logos, background images, headers and footers, and exact fonts, colours and spacing are not carried across.
- Table detection is based on column alignment. Tables without consistent columns, or with merged cells, may come through as ordinary paragraphs.
- Recognition accuracy on scans depends on the source. We have not measured it on your documents and do not publish an accuracy figure.