canivete
    en

    TOOL · DOCUMENTS

    Extract text from photos and PDFs (OCR)

    Photographed a document and need the text? Drop the photo or scanned PDF and take the editable text — read right in your browser, nothing sent anywhere.

    on the first read, the reading engine (~15 MB) is downloaded — after that it stays saved in your browser.

    OCR isn't magic: a tilted, dark or blurry photo makes it worse. Shoot straight and well lit and it improves a lot.

    Processed in your browser — your files never leave your computer.

    How it works

    1. Drop your photos or PDF here

      Drag up to 20 photos (JPG, PNG or WebP, up to 20 MB each) or 1 scanned PDF up to 50 MB. The PDF goes alone, without mixing in photos.

    2. Wait for the reading

      On the first run, the reading engine (~15 MB) is downloaded and stays saved in your browser. After that, each photo or page is read with a progress bar.

    3. Review, copy or download

      The text appears in an editable box — fix what you need, copy everything at once or download as .txt. Text from PDFs comes with page markers.

    Frequently asked questions

    Does it work with handwriting?

    Poorly — honesty matters here. OCR was trained on printed type: typed documents, books, receipts, signs. Handwritten text usually comes out wrong or turns into random characters. For handwriting, there is still no substitute for typing it out.

    Which languages does it read?

    Portuguese and English, even mixed in the same document — the engine loads both at once, with accents in place. Other languages may come out, but with no guarantee.

    Why does it download ~15 MB the first time?

    That is the reading engine (OCR) coming down to your browser — first time only; after that it stays saved. Note the direction: the engine comes to you, your files never go to it.

    Is my document uploaded to a server?

    No. The reading happens inside your browser, on your device. The only download is the engine; your photos and PDFs never leave your computer.

    Does the text come out perfect?

    No — OCR is not magic. Swapped accents and confused letters happen, especially with tilted, dark or blurry photos. A straight, well-lit photo improves things a lot; review the text before using it — the box is editable on purpose.

    Is there a page limit for PDFs?

    Yes: 20 pages per run, to keep your browser from freezing. Longer PDFs are read up to page 20, with a notice — split the PDF with the Split PDF tool and run again for the rest.

    Other tools