Check the source before converting
Try selecting a sentence in a PDF reader. If selectable characters are present, PDF to text can reuse the document’s text layer without recognizing every pixel. If selection only grabs an image, export pages with PDF to images and use Image to text. A scan may already have an invisible OCR layer; selection alone does not prove that layer is accurate.
Extract only what you need
Choose one PDF up to 20 MB and 100 pages. Leave the page field empty for the whole document, or enter a range such as 2-4,7. Page numbers refer to physical PDF pages, not printed chapter or Roman-numeral labels. Overlapping selections are deduplicated and returned in document order. Invalid selections stop the job rather than silently exporting a different range.
Review characters and reading order
Use the page selector to compare the result with the original. Check names, decimal points, minus signs and footnotes. Tables can lose column boundaries; two-column layouts can mix sentences. Copying extracted text into an email or a language model without reviewing it can transmit errors as well as sensitive content. Pages with no extracted text are listed explicitly.
Choose a useful reading copy
Download UTF-8 TXT with page separators when you need to trace statements back to the source. Turn separators off when you want a plain-text working copy, but retain the PDF for page references. The per-page preview is limited to 50,000 characters; the download contains the entire accepted extraction, subject to the one-million-character job limit.
Keep extraction separate from editing
The tool does not edit the PDF, reconstruct a Word document or make a scan searchable. It also does not retain your file after the tab is cleared or left. Save your own result and protect it like the source, especially when it contains names, contractual details or other private material. Use only documents you are allowed to copy.