Extract Text from an Image or PDF (OCR)

How to extract text from an image or PDF

1. Add your file. Drag a photo, screenshot or PDF onto the box, or click to browse. The tool reads it on your device.
2. Press Extract text. The OCR engine loads once, then reads the image (or every page of a PDF, one by one) and turns the printed characters into text.
3. Copy or download. The recognized text appears in a text box below. Copy it straight to your clipboard, or download it as a plain .txt file.

What OCR is actually useful for

OCR (optical character recognition) reads the shapes of printed letters in an image and turns them back into real, selectable text. The most common use is digitizing paperwork that only exists as a scan or a photo, an old contract, a form, a printed article, or a page from a book. Instead of retyping the whole thing by hand, OCR gets you most of the way there in seconds, and you just clean up any mistakes.

It is just as useful for everyday photos. Snap a picture of a receipt to pull the total and line items into an expense sheet. Photograph a whiteboard at the end of a meeting to save the notes as text instead of leaving them to fade on the board. Pull the text out of a screenshot, a slide, an infographic or a meme so you can quote it, translate it, or search for it later. Anywhere text exists only as pixels, OCR turns it back into words.

For PDFs specifically, this tool handles the case where a PDF is really just a stack of scanned page images with no real text layer underneath, common with older scanned documents and faxed paperwork. Each page is rendered and read in turn, and the text from every page is combined into one result with clear page markers, so you can tell exactly which paragraph came from which page.

Accuracy: what to expect

OCR works best on clean, printed text: a sharp scan of a typed document, a clear screenshot, or a well-lit photo of a printed page. Under those conditions accuracy is usually very high. Accuracy drops with anything that makes the letters harder to make out: low resolution photos, blurry or out-of-focus shots, a page photographed at an angle, poor lighting, low contrast between text and background, or unusual and decorative fonts. If your result looks garbled, try retaking the photo straight-on, in better light, and as close to the text as possible while keeping it in focus.

Handwriting is a different story. This tool, like most OCR engines, is trained and tuned for printed text, the kind that comes out of a printer or a typewriter. Cursive or handwritten notes use a completely different recognition problem, and results on handwriting are noticeably less reliable. It may pick up a word here and there, but treat handwritten output as a rough starting point to correct, not a finished transcript.

Why this runs in your browser

Instead of sending your file to a server to be read by someone else's software, the entire OCR model runs locally on your device using WebAssembly, a technology that lets full programs run at near-native speed directly inside the browser tab. The model files are downloaded once when you open this page, and after that every image or PDF you drop in is processed entirely on your machine. That matters most for anything sensitive, IDs, medical paperwork, financial statements, or contracts, since the content of the file never leaves your computer.

Frequently asked questions

Related: PDF to JPG / Images · compare PDF · text cleaner.