Step-by-step guide

How to Extract Text From a PDF

Updated October 9, 2026

If a PDF was typed, its text is already inside and can be read exactly and instantly. If it was scanned, each page is a picture, and the text has to be recognized with OCR (optical character recognition). FileICU's PDF to Text does both: it reads typed pages directly and runs OCR only on scanned pages, in English, French, Spanish, German, Arabic and Urdu.

Open PDF to Text

PDF to Text with a PDF chosen and the choice between plain text and a searchable PDF
PDF to Text on FileICU.

How to do it with PDF to Text

  1. Open PDF to Text and choose your PDF, typed or scanned, up to 25 MB.
  2. Choose plain text to copy or download the words, or a searchable PDF to keep the page images with a hidden text layer.
  3. Select Extract text. Pages are handled one at a time, and the language of scanned pages is detected automatically.
  4. Copy the text, download it as a .txt file, or download the searchable PDF.
  5. For scanned pages, read the result once: OCR can misread characters.

Typed PDF or scan: how to tell

Open the PDF and try to select a word. If the word highlights, the text is real text and will be extracted exactly. If the whole page highlights like a photo, or nothing happens, the page is an image and needs OCR.

Many files mix both, such as a typed report with a scanned signature page. PDF to Text handles each page the right way.

Example: a scanned letter

A two-page letter scanned at the office arrives as a PDF of two images. Choosing plain text gives you the letter's words, ready to paste into an email or document. Choosing a searchable PDF gives a file that looks exactly like the scan, but you can now search it for a name or copy a paragraph.

The first time you recognize a language, its OCR data is downloaded once from this website, about 0.7 to 3 MB, plus about 1.5 MB for the engine. Your PDF is never part of that download.

Getting better OCR results

OCR works best on clean, straight, high-resolution print. Faint copies, skewed photos of pages, small text and decorative fonts all reduce accuracy, and handwriting is not supported.

  • Scan at 300 DPI if you can, rather than photographing the page.
  • Keep the page straight and evenly lit.
  • Check names, numbers and amounts by eye after recognition.

Common problems

The extracted text comes out in the wrong order.
Multi-column layouts and text boxes can be read in a different order than you would read them. Copy the parts you need, or check them against the original.
Some characters are wrong after OCR.
The scan is probably low resolution or unclear. Rescan the page more clearly, or correct the few wrong characters by hand.
Arabic or Urdu text is not recognized well.
Clean Naskh print works best. Nastaliq, used in most printed Urdu, small text and low-contrast scans give less accurate results.
It is slow on my phone.
OCR uses a lot of memory and processing. Shorter documents, or a computer for long scans, work better.

Tools for this task