Try it

Image to Text Converter (OCR)

Pull the text out of a photo, screenshot or scanned PDF instead of retyping it. Optical character recognition runs right in your browser with the Tesseract engine, and you can copy the result or save it as a TXT file. Your image is never uploaded.

Text language

Files are processed right in your browser and never uploaded to a server. Open-source libraries

TL;DR. Add an image or a PDF of up to 20 pages, choose English or English + Russian, and get editable text to copy or download as TXT. Printed text works best; handwriting is recognized poorly.

How to extract text from an image

  1. Add a file

    Drop a JPG, PNG, WebP or HEIC image, or a PDF of up to 20 pages.

  2. Choose the language

    Use English + Russian (default) or English only. Picking only the languages in your document improves accuracy.

  3. Run recognition

    The first run downloads the language data once, about 10 MB. After that, recognition starts right away.

  4. Copy or save

    Check the text, then copy it to the clipboard or download it as a TXT file.

Recognition happens on your device

OCR runs inside your browser. On first use, the engine and its language models (about 10 MB) are downloaded once from the same site and cached. Your image or PDF itself is never sent anywhere.

You can verify it: open DevTools (F12 or Cmd+Option+I), switch to the Network tab and recognize a file. You will see the scripts and language files loading from the same domain, with no request carrying your image. On later runs, the models come from the browser cache.

How to get more accurate results

  • Use the highest-resolution image you have; tiny or blurry text causes most errors.
  • Crop away unrelated areas such as backgrounds, photos and decorations.
  • Keep the page straight and evenly lit, without glare or deep shadows.
  • Choose English only for English-only documents.
  • Always proofread numbers, names and addresses before using the text.

Where image to text saves time

OCR is most useful when text is locked inside a picture: a screenshot of a message or article, a photo of a printed notice, a scanned letter or a PDF made from scans where nothing can be selected. Instead of retyping it, you get editable text that you can paste into a document, an email, a translator or a search box.

HEIC photos taken on an iPhone can be added directly, without converting them to JPG first.

What OCR handles well and what it doesn’t

ContentResultTip
Printed text, screenshotsRecognized wellUse a sharp, well-lit image
Scanned PDF pagesRecognized wellUp to 20 pages per file
Photos of documentsGood if the page is flat and in focusShoot straight on, avoid shadows
TablesText is read line by lineCell structure is not preserved
HandwritingRecognized poorlyExpect errors; retype if accuracy matters

Image to text: frequently asked questions

Clean printed text and screenshots are usually recognized with few mistakes. Accuracy drops with blur, low resolution, unusual fonts or busy backgrounds. Always proofread important details like numbers and names.

Handwriting is recognized poorly, because the Tesseract engine is designed for printed text. You may get some words right, but expect many errors.

The text inside a table is recognized, but line by line, without the cell structure. You will need to rebuild columns yourself in a spreadsheet or editor.

You can choose English + Russian, which is the default, or English only. Selecting just the language of your document gives the best accuracy.

Yes, PDFs of up to 20 pages are supported. This is useful for scanned documents where text cannot be selected.

On first use, the browser downloads the language models, about 10 MB, from the same site. They are cached, so later runs start much faster.

Related tools

Protect the rest of your traffic too

LiMP VPN encrypts your connection in cafés, airports and at home. Your files already stay on your device — keep everything else private as well.

LiMP VPN features →