PDF to Word Converter
Turn a PDF — text-based, scanned, or a mix of both — or a JPG/PNG photo of a document into an editable Word (.docx) document without leaving your browser. Your file is never uploaded — reading, OCR, layout analysis and document assembly all happen on your own device.
Limits: up to 50 MB and 50 pages. Files are processed in your browser and never uploaded.
Your file is processed locally in your browser and is not uploaded to BurnJet. Nothing about the document (contents, name or metadata) is sent anywhere.
How to use it
- 1
Add your file
Drag a PDF, JPG or PNG onto the drop zone or browse for it. The file stays on your device.
- 2
Convert
BurnJet reads each page. Pages with real text are extracted directly; scanned or photographed pages are recognized on-device with OCR. Everything is rebuilt as one Word document in a background worker.
- 3
Download the .docx
Save the Word file and keep editing. Use "Convert another" to start over.
How the conversion works
- The PDF is parsed with pdf.js, page by page. Each page's own selectable text tells BurnJet whether it is a normal text page or a scanned/image page.
- Normal pages: text fragments are extracted with their position and font size, grouped into lines and paragraphs, and headings/bold/italic are detected from size and font.
- Scanned or image pages: the page is rendered to an image, lightly cleaned up (contrast, noise reduction, deskew), and recognized on-device with OCR (Tesseract) in a Web Worker so the tab stays responsive.
- JPG/PNG uploads go straight through the same OCR step as a single-page document.
- Every page — normal or OCR'd — is combined in the right order and sent to a Web Worker, which builds a real .docx file with the open-source "docx" library, including simple tables where they can be reliably detected.
Limitations & privacy
- Normal PDF pages convert most precisely; scanned pages, photos and JPG/PNG uploads depend on OCR, so recognition quality depends on scan/photo clarity.
- OCR does not detect bold, italic or precise alignment — recognized text is produced clean and editable rather than with guessed formatting.
- Simple tables are reconstructed as real Word tables only when they can be reliably detected; otherwise their text is kept in reading order.
- Exact fonts, colours, spacing and page breaks from the original are approximated, not preserved 1:1.
- Very large, very dense, or heavily scanned documents may take longer and can hit the size/page limit or the safety timeout.
- The file, its name and its contents never leave your browser and are not written to any database or storage.
Frequently asked questions
Is my PDF or image uploaded to a server?
No. Everything — reading the file, extracting text, and running OCR on scanned pages — happens entirely in your browser using WebAssembly and Web Workers. No network request carries the file, its name or its contents, and nothing is stored.
Which files work best?
PDFs created from a word processor or export pipeline, containing real, selectable text, convert most precisely. Scanned PDFs and JPG/PNG photos of documents are also supported through on-device OCR, though recognized text depends on scan/photo quality.
Can it convert a scanned PDF?
Yes. Pages with little or no selectable text are automatically rendered and run through on-device OCR (Tesseract), then turned into real, editable text — never just a picture of the page. A PDF that mixes normal and scanned pages is handled correctly: each page is processed the way it needs to be, in the right order.
Can I upload a JPG or PNG instead of a PDF?
Yes. Uploading a JPG or PNG photo or scan of a document runs it through the same on-device OCR pipeline and produces a one-page editable Word document.
Does it support languages other than English?
For scanned pages and images, choose an OCR Language next to the Convert button: English, Hindi, or English + Hindi. This only affects OCR — it has no effect on PDFs with normal selectable text.
What about tables, columns and images?
Paragraphs, heading levels, and simple tables (like those in invoices, forms and statements) are reconstructed where they can be reliably detected, and embedded images are carried over on normal PDF pages. When formatting can't be reliably detected — which is more likely for OCR'd text — BurnJet produces clean, correctly-worded editable text rather than guessing at broken formatting.
My PDF is password-protected. Can I still convert it?
Yes, if you know the password. Enter it when prompted and the conversion continues locally. BurnJet never transmits or stores the password.
Is there a size or page limit?
Yes. Limits are set by the site administrator and capped by a hard safety ceiling. The current effective limits are shown above the drop zone.