MultiKit

PDF to Word (DOCX): Text, Headings and Pictures, in Your Browser

  • Free
  • No sign-up
  • Runs in your browser
  • Version 1.0.2
  • Last updated 2026-10-25
Report a problem

Get the text of a PDF into a Word document you can edit, without sending the file anywhere. It works best on documents of paragraphs, headings, lists and pictures; tables and complex layouts are not kept as they were.

Works best on documents of paragraphs, headings, lists and pictures. Tables come out as lines of text with tabs between the cells, a page in two columns is read down each column, pages with three or more columns are read across, and fonts, colors and exact positions are not kept. Check the result and fix what needs fixing in Word.

    Everything you enter is processed in your browser and is never sent to a server.

    Browsers:Chrome119+Edge119+Firefox121+Safari17.4+Opera105+Samsung Internet25.0+Browsers

    How to use

    1. Choose or drop one or more PDF files. A PDF with a password has to be unlocked first.
    2. Leave Keep the pictures on to carry the pictures over. Leave Leave out page numbers and headers on to drop page numbers and headers that repeat on every page. Leave Read scanned pages with OCR on, and pick the language, for pages that are pictures of text.
    3. Press Convert to Word. Each file is read page by page, and the page you are on is shown. You can stop at any time.
    4. Download the .docx file and open it in Word, LibreOffice or Google Docs. Check it: headings, paragraphs, lists and pictures are rebuilt from the PDF, but tables, columns and exact layout are not.

    Examples

    A report with a title, section headings, paragraphs and a bulleted list
    → A Word file with a Title and Heading styles, normal paragraphs and a bulleted list
    The headings are found from the size of the text. The paragraphs are found from the spacing and indents of the lines.
    A document with a price table
    → Each table row becomes a line of text, with tabs between the cells
    The table is not rebuilt as a Word table. You can select the lines and use Convert Text to Table in Word.
    A scanned letter
    → The text read by OCR, in paragraphs
    OCR is a best guess, so read it through. Headings and fonts of a scan cannot be found.

    How it works

    A .docx file is a ZIP archive of XML files, and writing one is not hard. The hard part is reading a PDF the right way. A PDF holds text as pieces placed at exact positions on a page, each with a font and a size, and it has no idea of a paragraph, a heading, a list or a table. This tool reads those pieces with pdf.js and rebuilds a structure from where they are.

    Pieces on the same baseline become a line. The size most of the text is set in is the body size. Lines that are clearly larger, and short, are headings: the largest size is the first level, the next the second, up to three levels. Consecutive lines join into a paragraph while the spacing between them is that of a paragraph, and a new paragraph starts at a bigger gap or at a line pushed in from the margin. Lines that start with a bullet become list items. A word split by a hyphen at the end of a line is joined again. Page numbers and headers that repeat on most pages are left out. The pictures painted on a page are taken out and placed between the text where they sit.

    The Word file is written with the docx library, with each PDF page starting on a new page of the document and the size of the first PDF page. For a page that has no text, the tool draws it and reads it with the Tesseract OCR engine, the same one used by the Image to Text tool.

    How well it works was measured in a real browser with a 3-page report printed to PDF by Chrome: all 767 words of the text were kept, the title and the six section headings were found, the picture was carried over, the running header and the page numbers were left out, each row of a price table came out as a line with tabs, and the seven list items came out one to a line. A scan with a title and four lines of text was read by OCR with all 33 words right. A page set in two columns kept all 118 words and came out in reading order, down the left column and then the right: every pair of neighbouring words of the source was still neighbours. These are test files made for this page; real documents vary.

    What it does not do: it does not rebuild tables, it follows two columns but not three or more, and it does not keep fonts, colors, text boxes, forms or footnotes. Check the result before you use it.

    When is PDF to Word the right choice?

    • Best fit: reports, letters and articles whose main value is editable text, headings, lists and pictures.
    • Input and output: one or more PDF files become editable DOCX files. Image-only pages can be read with OCR.
    • Privacy and quality: conversion and OCR run locally. Digital PDFs are usually more accurate than scans; names, numbers and line breaks still need review.
    • Known limit: choose a layout-preserving PDF editor instead when exact tables, forms, fonts, footnotes or three-column pages must stay unchanged. This converter favors editable structure over pixel-perfect layout.

    Sources

    Frequently asked questions

    How do I convert a PDF to a Word document?
    Add the PDF and press Convert to Word, then download the .docx file. The tool reads where each piece of text sits on every page, works out the headings, paragraphs and lists, and writes a Word file from them.
    Will the layout be the same as in the PDF?
    No. A PDF only says where each piece of text goes on a page. It does not say what a heading, a paragraph or a table is, so this tool guesses. Headings, paragraphs, bullet lists and pictures usually come out well. Tables become lines with tabs, a page in two columns is read down the left column and then the right, pages with three or more columns are read across, and fonts, colors and exact positions are not kept.
    Can it convert a scanned PDF?
    Yes. Pages that have no text are drawn and read with OCR in the language you choose, then written as paragraphs. OCR can make mistakes, and the headings and fonts of a scan are not found.
    What happens to tables and columns?
    A table is not rebuilt. Each row is written as a line with tabs between the cells, which you can turn into a table in Word with Convert Text to Table. A page set in two columns is read down the left column and then down the right one, with a title across both columns kept above them. A page with three or more columns, or with columns that are not evenly spaced, is read line by line across the page, so the columns can run together.
    Are bullet and numbered lists kept?
    A line that starts with a bullet mark such as • or a dash becomes a bullet in the Word file. Numbered lines stay as text with their numbers, one item to a line. Some programs draw the bullet as a shape instead of a character, as Chrome does when it prints a web page to PDF, and then there is no mark to find: the items still come out one to a line, without a bullet.
    Are the pictures kept?
    Yes, the pictures painted on a page are placed in the Word file between the text, at the size they had, and a picture that fills the whole page is left out. Their exact position beside text is not kept.
    Is my PDF uploaded?
    No. The PDF is read and the Word file is made in your browser. OCR runs in your browser too, with its data served from this site.
    Why are some words joined or split wrongly?
    A line that ends in a hyphen inside a word is joined to the next line, and Chinese and Japanese lines are joined without spaces. Some PDFs set each letter on its own or use unusual spacing, and then words can come out wrong. Check the result.
    Version history
    • 1.0.2 — Adds a practical decision guide covering best-fit uses, inputs and outputs, privacy, quality trade-offs and known limits.
    • 1.0.1 — Makes the search title and description state the supported task, key formats and privacy benefit more clearly.
    • 1.0.0 — First release.

    Last updated: