How to Convert a PDF to Plain Text (TXT)
For piping into ChatGPT, building a search index, or running NLP scripts, plain text is what you want — not PDF. PDFShed extracts the text in one step.
The problem
You have a 200-page report you want to feed to an LLM for summarization. You need plain text — not the PDF, not Word, not even Markdown. PDFShed extracts cleanly.
Step-by-step
- 1
Open the PDF to Text tool
Drop your PDF in.
- 2
Choose extraction mode
"Reading order" preserves natural top-to-bottom, left-to-right flow. "Layout-preserving" keeps columns and tabs (useful for code or formatted output).
- 3
Extract
Each page's text is concatenated into a single .txt file.
- 4
Download
Open in any text editor, paste into ChatGPT, or pipe through scripts.
- 5
For scanned PDFs, OCR first
Image-only PDFs need [OCR](/en/tools/ocr-pdf) first. Skip this step for native text PDFs.
Pro tips
- •For LLM context, "Reading order" is what you want. Models prefer narrative-flow text.
- •For code-bearing PDFs (technical docs), "Layout-preserving" maintains indentation.
- •Tables don't survive text extraction cleanly. For tables, use [PDF to Excel](/en/tools/pdf-to-excel) instead.
- •After extraction, you can re-format or chunk the text for downstream processing.
Frequently asked questions
Will images and charts come through?
No — text only. Images, charts, and graphics are dropped. For full visual content, stay with PDF or convert to PowerPoint.
How does this differ from PDF to Word?
PDF to Text strips all formatting (no fonts, sizes, italics, bullets). PDF to Word preserves formatting. For LLM input, plain text is usually better.
What about scanned PDFs?
Run [OCR PDF](/en/tools/ocr-pdf) first. Without OCR, scanned PDFs return empty text because the "text" is image data.
Will it preserve page breaks?
A blank line is inserted between pages by default. Optional "page marker" mode adds [PAGE 1], [PAGE 2] headers for downstream parsing.