How to convert an image to text (OCR) for free

Got text trapped in a screenshot or photo? Here's how to pull it out into editable text for free with on-device OCR — no upload, no account, and your image never leaves your browser.

1. Open the Image to Text tool. Open the free Image to Text (OCR) tool — OCR runs on your device.
2. Add your image. Drop in a screenshot or photo (JPG, PNG, etc.). It's processed locally.
3. Run OCR. Let the tool detect and extract the text from the image.
4. Copy or download. Copy the recognized text or save it — ready to paste anywhere.

What OCR reads well, and what it does not

Optical character recognition looks for the shapes of printed letters. It is reliable on typed or printed text: screenshots, scanned documents, book pages, receipts, labels, slides. It struggles with handwriting, decorative or very thin fonts, text on curved surfaces, low light, heavy shadows and photos taken at a steep angle. The Image to Text tool runs recognition on your device and works best with common Latin-script languages.

Prepare the image first

• Shoot straight on. Hold the phone parallel to the page. Skewed lines are the most common cause of garbage output.
• Get even light. Avoid a shadow from your hand or phone across the text. Daylight from a window beats a ceiling lamp.
• Crop to the text. Backgrounds, photos and logos around the text add noise; crop them out before running OCR.
• Use the highest resolution you have. A screenshot at 100% zoom or a full-resolution photo reads better than a thumbnail or a compressed forward from a chat app.
• Increase contrast if the scan is grey. Dark text on a light background is what OCR expects.

Check these after extraction

OCR gets words right far more often than it gets characters right in isolation, so the errors cluster in numbers and names. Common swaps: 0 and O, 1, l and I, 5 and S, 8 and B, rn and m. If you are extracting totals from a receipt, an invoice number or a phone number, verify them against the image before you use them. Tables and multi-column layouts come out line by line; crop each column and run it separately to keep the order.

Typical jobs

• Receipts and invoices into a spreadsheet for expenses or bookkeeping.
• Quotes from books and printed reports without retyping them.
• Whiteboard photos from a meeting into editable notes.
• Error messages and code in screenshots that cannot be copied as text.
• Scanned PDFs: for a multi-page scan use OCR PDF, which keeps the pages together; for a PDF that already contains text, PDF to Word is the faster route.

Privacy

Recognition happens in your browser. The image is never uploaded, and the extracted text stays on your device until you copy or download it. That makes the tool suitable for documents you would not want on a third-party server, such as ID cards, contracts and medical letters.

When the volume gets serious

Reading one receipt is a job for this tool. Reading three hundred supplier invoices a month, matching them to orders and posting the totals to your accounting system is a job for AI workflow automation: extraction, validation and routing that run without anyone retyping.

FAQ

What languages does the OCR support?

It handles standard Latin-script text well; clear, high-contrast images give the most accurate results.

Is my image uploaded?

No — OCR runs in your browser, so the image and the extracted text stay on your device.

Can it read handwriting?

OCR works best on printed or typed text; handwriting is hit-or-miss.

Can it extract text from a PDF?

For a scanned PDF use the OCR PDF tool, which processes every page. For a PDF that already contains text, PDF to Word or copy-and-paste is quicker.

Why is the output jumbled?

Usually a multi-column or table layout: OCR reads across the page line by line. Crop each column or cell and run it separately.

Does it keep the formatting?

No. You get plain text, which is what you want for pasting into a document or spreadsheet. Re-apply headings and bold afterwards.

Does it work on photos of screens?

Often, yes, if the photo is sharp and straight. A screenshot taken on the device itself is always better than a photo of a screen.

⚡ Extracting data from invoices, receipts or forms every week? AI workflow automation reads, checks and routes them without anyone retyping →