PDF to Markdown
Extract the text layer into a clean .md with paragraphs and heading levels
Drag & drop files here
or browse files
Text-layer PDFs only - run OCR on scans first
Honest scope: PDFs carry no semantic structure, so paragraphs and headings are rebuilt from spacing and font size - text-dominant documents work best; two columns are read column by column; tables and images are not included.
About this tool
Moving papers, docs and manuals into Notion, Obsidian or Yuque works best through Markdown. This tool extracts the PDF text layer and rebuilds paragraphs from text coordinates using line spacing and indents, producing a clean .md - not one hard line break per visual line, but real flowing text ready to paste into note apps.
Honest scope: PDF has no semantic structure, so reassembly is geometric - text-dominant documents work best; two-column layouts are detected and read column by column; scanned PDFs have no text layer, so run OCR first. Tables and images do not enter the output (PDFs carry no table structure; images need other tools).
The result appears in an editable box - fix heading levels and merge broken paragraphs, then download. Everything runs locally; files never upload.
Features
- Text-layer extraction with geometric paragraph reassembly
- Two-column detection for correct reading order
- Heading levels inferred from font size
- Editable result on the page before download
- Text-layer PDFs only - run OCR first for scans
How to use
- Upload a text-layer PDF (OCR scans first)
- Pages are extracted and reassembled with character stats
- Review and fine-tune the Markdown in the editor
- Download the .md file
Frequently asked questions
- Why doesn’t the output look like the original layout?
- Markdown targets structure, not appearance: paragraphs reassemble by spacing and headings by font size, while fonts, colors and running heads do not survive. For visual fidelity use PDF to Word or PDF to Images.
- Are two-column papers handled correctly?
- Yes - the column layout is detected from text coordinates and the left column is read before the right, preserving order. Three-column spreads and cross-column figures can still confuse the detector, so spot-check the sequence after converting.
- Why is a scanned PDF’s output empty?
- Scanned pages are images with no text layer to extract. Run the site’s OCR tool first to convert the image into selectable text, then convert - or paste the OCR result straight into your notes.