Open a scanned document and try to select a sentence with your cursor. If nothing highlights, you're looking at an image-only PDF: the page is a picture, and the letters you see are just colored pixels arranged to resemble words. Your device has no idea there is any text there at all. It can display the page and print it, but it can't find a phrase, copy a paragraph, or read the words aloud, because as far as the software is concerned there are no words, only an image. Making that file searchable means adding a text layer underneath the image: the picture stays exactly as it looks, but a machine-readable copy of every word sits invisibly behind it, aligned to where each word appears. That's what optical character recognition, or OCR, produces.

Image-only versus searchable: why the difference matters

The gap between the two shows up the moment you need to do anything beyond looking at the page. With an image-only scan you can't search for a name, you can't copy an address into an email without retyping it, and a screen reader has nothing to announce. A searchable PDF fixes all of that while looking exactly the same on screen, which is why adding a text layer is almost always worth the couple of minutes it takes. Once the layer is in place, the same file suddenly supports the everyday things you'd expect a document to do.

  • Search: press find and jump straight to the clause, figure, or name you need instead of scrolling through pages.
  • Copy: pull a quote or an address out cleanly rather than retyping it by hand and risking typos.
  • Accessibility: assistive tools can announce the text for readers who rely on them.
  • Reuse: feed the recognized words into a summary, a spreadsheet, or a translation later on.

Prep the scan before you run OCR

OCR accuracy depends heavily on what you feed it, and a few minutes of tidying up front saves a lot of cleanup afterward. A crooked, cluttered, or upside-down page confuses the recognizer, so it helps to straighten and clean each page first. If a page came in sideways or upside down, use Rotate to set every page upright, since OCR reads left to right and expects text the right way up. Cropping away scanner shadows, punch holes, and the dark border keeps the recognizer focused only on the words. And a clean, high-contrast scan, dark text on a plain light background, reads far better than a faint or grayish copy. If you're capturing pages fresh from paper, Scan to PDF helps you get square, evenly lit shots to begin with.

Why straightening matters more than you'd think

Deskewing matters more than most people expect. Even a few degrees of tilt can turn clean sentences into scrambled characters, because the software tries to read across slanted rows and quickly loses the line. Straight, upright pages give you far fewer surprises, so it's worth glancing at each page and squaring it before you go any further. The same goes for contrast and clutter: the recognizer is essentially guessing at letter shapes, and anything that muddies those shapes, a shadow, a coffee stain, a faint carbon copy, is one more thing it can misread. Clean input is the single biggest lever you have over the quality of the result, far more than any setting.

Running OCR in your browser

The OCR tool runs entirely on your own device using WebAssembly, so your scanned pages never leave the browser. There's no upload, no account, and no watermark, which is worth knowing when the document is a signed agreement, a medical form, or anything else you'd rather not hand to a server. Point OCR at your prepped file and it works through each page, detecting characters and building the hidden text layer as it goes. Larger documents take longer, since every page is analyzed as an image, but when it finishes you have a PDF that looks the same and is now fully searchable and selectable. A quick way to confirm it worked is to search for a word you know is on the page; if it jumps to the right spot, the layer is doing its job.

Getting the text back out

Sometimes you don't want a searchable PDF at all; you want the raw words in a plain form you can paste elsewhere. Once a document has a text layer, PDF to Text lifts the words out into a clean text file, ready to drop into notes, a spreadsheet, or an email. Expect occasional slips on handwriting, unusual fonts, or faint print, since recognition is strong but never perfect, and check the spacing on tables and forms, where columns can merge together in plain text. Put together, the whole flow is simple and repeatable: straighten and clean the scan with Rotate and cropping, capture fresh pages squarely with Scan to PDF if you're starting from paper, add the invisible text layer with OCR, then either keep the searchable PDF or pull the words out with PDF to Text. A pile of flat photographs becomes a document you can actually search, quote, and reuse, without any of it ever leaving your machine.