How to convert scanned PDFs, Word and tables to Markdown

How to convert to high-fidelity Markdown: the short answer

Pick the right converter (PDF, Word or Markdown to PDF), click "Open converter", upload the file and wait: when it finishes, you download Markdown with headings, bold, lists and tables rebuilt, or the formatted PDF. A scanned PDF goes through text recognition (OCR) automatically. The conversion happens on our server and the file is discarded afterwards.

What high-fidelity Markdown conversion means

Markdown is plain text with light marks: one symbol signals a heading, another a list, and so on. A high-fidelity conversion doesn't just dump the text: it rebuilds the document's structure. It recognizes that a line is a heading, that a block is a table, that a snippet is code and, for a scanned PDF, reads the text trapped inside the image before organizing it. The result is Markdown you can edit and publish without spending hours rebuilding tables and putting headings back by hand.

When to use the high-fidelity version (and when not to)

• A scanned PDF or a photo of a document, where the text is trapped in the image and can't be copied.
• Tables with several columns, which a simple conversion tends to scramble or flatten into a single line.
• Long Word documents, with headings, lists, links and tables you want to keep without copying and pasting.
• Code snippets that need to come out in blocks, without losing indentation or turning into flowing text.
• Producing a clean, presentable PDF from a README or from notes written in Markdown.

If your PDF already has selectable text and you just want something fast and fully in the browser, the simple PDF-to-Markdown version does it without sending anything to a server. The high-fidelity version is for scanned PDFs, complex tables and code, the cases where the quick conversion isn't enough.

Step by step to convert to high-fidelity Markdown

1. Pick the direction: PDF to Markdown (v2), Word to Markdown (v2) or Markdown to PDF (v2). They're three separate converters, one per path.
2. Click "Open converter". The message "Preparing the converter… it may take a few seconds" appears; that initial wait doesn't depend on the file size.
3. Upload the file in the converter window: only .pdf in the first, only .docx in the second (the old .doc isn't accepted) and only an .md file in the third (there's no box to paste text). Limit: 200 MB.
4. Wait while the server reads the structure and click "Baixar MD" (or "Baixar PDF" on the return trip) to get the finished file; the converter screen is in Portuguese.

PDF to Markdown: what comes out and how a scan is read

The PDF converter rebuilds headings, bold and italic, lists and tables from the position and style of the text on the page. When the page is an image (scanned or photographed), it triggers text recognition (OCR) on its own and returns the text it read, already organized. That reading uses the English dictionary; in Portuguese or Spanish text, accents and cedillas may come out wrong, so check the result. As with any image reading, accuracy depends on how sharp the original is: a well-scanned document comes out nearly perfect; a crooked photo with a shadow may bring letter errors worth a check. If you only need the text of the scan, without structure, the in-browser OCR is enough: see how to extract text from a scanned PDF with OCR.

Word to Markdown: headings, lists, links and tables (without the images)

Word (.docx) to Markdown has its own converter, Word → Markdown v2, ideal for pulling the text out of a report without copying and pasting page by page: headings, bold, lists, links and tables come out in place. One limitation you need to know first: the images embedded in the .docx aren't included in the downloaded .md, which is a single text file; references to them may point to a path that doesn't exist. If the report has important figures, keep them separately.

There's also the quick version, Convert Word to Markdown, which runs entirely in your browser and already covers headings, lists, bold, tables and links. Start there; the v2 is the fallback for long or complex documents where the quick version scrambles something. It is explained in how to convert documents between Word, EPUB, Markdown and HTML.

Markdown to PDF: the return trip

When you already have the Markdown ready (a README, some documentation, a note), the Markdown to PDF (v2) converter produces a well-formatted PDF, with headings, lists, tables and code blocks with syntax highlighting (color-coded code), ready to send to someone who doesn't use Markdown. It expects an .md file; if your text lives in an editor, save it as .md first. Images referenced by a local path (a figure saved next to the .md) don't make it into the PDF, because only the .md is sent.

For simple notes, the quick Markdown to PDF version produces the file right in your browser, with headings, lists, tables, code and emphasis, but without syntax highlighting or LibreOffice's layout polish. Its step-by-step instructions are in how to convert TXT, Markdown or HTML to PDF.

Common problems and limits

• "Preparing the converter" takes a while: the service may be asleep and needs a few seconds to wake up. It isn't the file; wait.
• File refused: each converter accepts a single extension (.pdf, .docx or .md). An old .doc must be saved as .docx in Word first.
• File too large: the limit is 200 MB per conversion.
• Scanned text with errors: improve the original (sharper, no shadow, straight page) and convert again.
• Images missing from the Markdown made from Word: that's the known limitation; figures aren't included.

Unlike most AtlasDocs tools, which run in your browser, these three high-fidelity conversions happen on our server; that's what makes it possible to read scanned PDFs and build tables and code accurately. The file is used only to produce the result, in a temporary folder, and isn't stored.