GUIDES · PDF TO MARKDOWN
Turn a PDF into Markdown you can check against the page.
Papers, textbooks and standards are hard to convert well: equations turn into garbled symbols, tables lose their columns, and a scan has no text at all. Docr’d rebuilds each page into structured blocks, lets you compare every formula and table with the original crop, and exports the result as Markdown with LaTeX math and the figures as image files.
What you get
The Markdown export is a ZIP file holding:
document.md, the document in reading order with its headings, paragraphs and lists.- Equations as LaTeX math:
$…$inline and$$…$$for display equations, with\tag{…}where the original was numbered, so the file renders in any Markdown viewer that supports math. - Tables as HTML tables inside the Markdown, so merged cells survive. Markdown’s own pipe tables cannot hold them. Some converters, pandoc included, drop raw HTML when writing other formats, so if you need the tables in LaTeX, use the LaTeX export instead.
- An
assetsfolder with each figure as an image file, linked from the text. document.json, the same content as structured blocks with their page positions, for scripts and pipelines.
Corrections you made in Docr’d are part of the export, and document.json marks the content as derived from the original PDF.
Convert a PDF
- Create an account and open the web app. Hosted OCR runs there, in Docr’d Cloud, never on your computer. Adding documents and OCR need cloud access: an activated free place, while one is available, or a paid plan. Your account shows which you have.
- Add the PDF to your library. It opens right away. A born-digital PDF is already readable and searchable; a scan has no text until it is reconstructed.
- Run reconstruction on the document and choose a quality. On a free place you can choose a page range, so a long book can go a chapter at a time.
- Open the document’s menu on its record bar and choose Export → Markdown. When the ZIP is ready, choose Download.
The same menu offers LaTeX, the OCR engine’s Native output exactly as it produced it, and an Annotated PDF with your marks drawn in. To grab one part instead of the whole document, the reconstruction editor can copy a section as Markdown.
Choosing a quality
| Fast | Text only. Good for plain prose; it does not rebuild layout, tables or formulas. |
|---|---|
| Balanced | Layout, headings, tables and formulas. The right choice for most papers and textbooks. |
| Max | The most thorough: adds charts, formula numbers and tables that continue across pages. Each page counts three times against your allowance. |
If you need equations or tables in your Markdown, choose Balanced or Max. Born-digital PDFs benefit too: their text is already there, and reconstruction adds the structured formulas, tables and headings.
Check before you export
No converter reads every page perfectly, so Docr’d keeps the evidence beside the result. Every block remembers where it came from on the page, and one click jumps back to that spot. Open a formula or table to see the crop of the original next to the rebuilt version, then fix the LaTeX or a cell in place. Fixes carry through to search and to later Markdown and LaTeX exports; Native output always stays exactly as the engine produced it.
Reconstructed text is searchable across your whole library, formulas included: type part of an equation, or search its LaTeX, to find where it appears.
What it costs
Reading, correcting, searching and exporting pages you have already reconstructed uses no OCR pages. Reconstruction does, and correcting in the web app needs cloud access:
- Free places include 100 Balanced pages a month while places are available. There are 50, and the shared free budget can pause free OCR until the next month.
- Paid plans include monthly OCR pages at any quality, with page packs for big projects and never an automatic overage charge. See pricing.
Docr’d is in early access, so you may run into rough edges. Tell us at support@getdocrd.com when a page converts badly; please leave out private documents.
When another tool fits better
Docr’d is built for people who keep and reread what they convert: the Markdown stays linked to its source page inside a research library with notes, citations and search. If you only need a single born-digital PDF converted once on your own machine, an open-source command-line converter such as Marker may suit you better. Docr’d does not read handwriting.