Scanned documents are just pictures of text: you can't search, select or copy them. This tool reads every page with OCR and adds an invisible text layer, so the PDF looks exactly the same but becomes searchable and copyable. You can also download the recognized text.
Last updated:
Recognized words
Advertisement
Advertisement
How to use the OCR PDF
Choose your scanned PDF by dropping it into the box or clicking to browse.
Pick the document language (and a second one if the text mixes languages).
Choose the accuracy. Balanced works for most scans; use 300 dpi for small print.
Press Start OCR. The first run downloads the OCR engine and language data, then each page takes a few seconds.
Download the searchable PDF, or download or copy the extracted text.
How a searchable PDF works
A scanned PDF stores each page as a picture. After OCR, the page has two layers: the original image you see, and underneath it a layer of text drawn in invisible mode (PDF text rendering mode 3). Each recognized word is placed and stretched to cover the printed word, so when you search or drag to select, the highlight lands on the right spot. A ToUnicode map tells PDF readers which characters the text represents, so copied text comes out correctly in any language.
Getting the best OCR results
Problem
What helps
Small print, footnotes, tables
Use the 300 dpi accuracy setting.
Mixed languages (e.g. English and French)
Pick both languages; the engine considers both dictionaries.
Photos of pages taken with a phone
Shoot straight from above in good light; avoid shadows and curved pages.
Faint or gray scans
Rescan in black and white or grayscale at 300 dpi with higher contrast.
Handwriting
OCR engines are trained on printed text; handwriting is recognized poorly.
Accuracy settings explained
Before recognition, each page is rendered to an image. A higher resolution gives the engine more pixels per letter, which helps with small text but takes longer. As a rule of thumb, OCR engines work best when capital letters are at least about 20 pixels tall. In 10-point text, capitals are roughly 20 pixels tall at 200 dpi and 30 pixels tall at 300 dpi.
Frequently asked questions
What does OCR PDF do?
OCR (optical character recognition) finds the letters in the images of a scanned PDF. This tool places the recognized words as invisible text exactly over the printed words, so the PDF looks the same but you can search it, select text and copy it.
How accurate is the OCR?
Clean, straight scans of printed text at 200 to 300 dpi usually reach 95 to 99 percent word accuracy. Accuracy drops with blurry or skewed photos, handwriting, very small print, unusual fonts and low contrast. The preview colors each word by confidence so you can spot doubtful areas.
Which languages are supported?
More than 70, including English, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, Chinese, Japanese and Korean. Choose the language the document is written in; selecting the wrong language noticeably lowers accuracy. Language data downloads once and is cached by your browser.
Will OCR change how my PDF looks or reduce quality?
No. The original page images are kept untouched and the text layer is invisible. The file size grows only slightly, usually by a few kilobytes per page.
Why are some pages skipped?
With “Skip pages that already have selectable text” turned on, pages that already contain real text (for example pages typed in Word) are not OCRed again, because their text is already searchable and more accurate than OCR. Untick the option to OCR every page.
Is my document uploaded to a server?
No. The PDF is rendered and recognized in your browser with pdf.js and Tesseract. Only the OCR engine and language data are downloaded; your document never leaves your device.
How long does OCR take?
Usually 2 to 10 seconds per page on a modern computer, depending on the accuracy setting and how much text the page has. The first run takes longer because the engine and language data are downloaded.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.