PDF to Word turns a PDF that carries a text layer into a .docx you can keep editing in Word or WPS Office: it reads the text coordinates on each page, joins character blocks into lines, then merges lines into paragraphs using three signals — a sudden jump in line spacing, a font-size change and a first-line indent — keeping an approximate size and bold state per paragraph. The goal is to get the text out together with its paragraph structure, not to reproduce the layout.

It is currently in Beta, with the paragraph rules still being tuned. Unlike Extract text, it outputs a .docx and tries to segment it properly; unlike PDF OCR it does no OCR at all — a scanned file yields almost no text here, and the page says so and offers a “Go to PDF OCR” button.

1. 01Upload a PDF— drag in a single text-based PDF; the page count appears.
2. 02Choose the pagination— the “Page breaks as in the original PDF” switch is on by default; turn it off to let paragraphs run on.
3. 03Start the conversion— the progress bar shows “Extracting text from page x/n” and then “Building the Word document…”; if fewer than 30 characters are found in the whole file, the tool stops and warns “Almost no text was extracted — this PDF may be a scan”.
4. 04Download— the .docx downloads under the original name, with “n pages · n paragraphs · about n characters” shown and a Download again button.

Beta

— paragraph inference and the output format may still change; check important documents paragraph by paragraph.

Kept

— text, paragraphs, an approximate size per paragraph (average line height), bold (font names containing bold, black or heavy) and optional page breaks; images, tables, columns, headers and footers, footnotes, hyperlinks, colors and the original fonts are not kept.

Paragraph rules

— line spacing above 1.75× the line height, a font-size change over 20% and a first-line indent; poetry, centered headings and lists can be merged or split incorrectly.

Columns and tables

— column text is stitched together by vertical position; tables become line after line of text, so use PDF to Excel when you need a table.

What it does not do

— it does not recognize scans (use PDF OCR first, which can also export DOCX), does not rebuild the layout and does not open encrypted PDFs (use Unlock PDF first).

The PDF is parsed, its text extracted and the .docx built in your browser; the file contents are never uploaded to any server, and the data is released from memory when you close the page.