Free Image to Text Converter: Extract Text from Photos, Scans and PDFs

Yes. Every tool is free, with no account, no sign up, no watermark and no limit on how many files you scan.

No. The OCR engine loads in the page the first time you use it. It works in current versions of Chrome, Edge, Safari and Firefox on desktop and mobile.

Yes. Every tool works on a phone screen, and the “Take a photo” buttons open your camera directly.

Yes. It is built for everyday paperwork: contact cards from events, receipts for expense reports, letters and forms, and screenshots shared in support tickets.

No. Batches are capped at 40 images or pages at a time to keep your browser responsive, but you can run as many batches as you like.

No. Images, PDFs and the extracted text are processed inside your browser with JavaScript and WebAssembly. Nothing you scan is sent anywhere.

The first time you use OCR, your browser downloads the open-source engine and the data file for your language from the jsDelivr CDN. Your files are not part of that request. After that, the browser usually serves them from its cache.

No. Refresh or close the tab and everything is gone. The only thing remembered is your preferred OCR language, stored in your own browser.

Yes. Because the file never leaves your device, it is no riskier than opening the photo in a local image viewer.

Tesseract, the open-source engine originally developed at HP and later Google, compiled to WebAssembly by the Tesseract.js project, with Scanly's own image enhancement and result checking on top.

JPG, PNG, WEBP, GIF, BMP, HEIC and HEIF photos from iPhones, and PDF files. Each PDF page is turned into an image and read like a scan.

Yes. Copy a screenshot or image and press Ctrl+V (Cmd+V on Mac) anywhere on the page. It goes to the tool you used last.

Yes. Add up to 40 images or PDF pages, then choose “Extract all”. Switch between pages with the thumbnails, or tick “Show all pages together” to copy or download everything in one go.

Yes. Drag a box across the preview and only that area is read. Use “Reset crop” to go back to the whole image.

Switch to the Layout view. It places every word according to where it sits on the page, so tables and columns stay aligned. You can download that version as a separate text file.

Yes. The Text view is fully editable, and your edits are what gets copied and downloaded.

Extract the text, open Download and choose “Searchable PDF”. With several pages, they are combined into one PDF.

Neat block handwriting sometimes works. Cursive and messy handwriting usually does not, because the engine is trained on printed text.

Yes. “Read aloud” uses your device's built-in speech voices and picks one that matches the OCR language when available.

55 languages, including English, Spanish, French, German, Portuguese, Italian, Dutch, Polish, Turkish, Russian, Ukrainian, Arabic, Urdu, Persian, Hebrew, Hindi, Bengali, Punjabi, Tamil, Telugu, Marathi, Chinese (Simplified and Traditional), Japanese, Korean, Thai, Vietnamese and Indonesian.

Yes. Choose a second language and both models are used together. This is ideal for English mixed with Arabic, Urdu, Hindi, Chinese and so on.

It is the engine's average confidence across every character it read, weighted by word length. Above 90 is usually near-perfect. Below 70 means you should review the highlighted words.

Usually the language does not match the text, the text is too small or blurry, or the image is mostly not text. Check the Quality signals panel, which tells you which of these applies.

Auto mode first reads a sharpened grayscale version. If confidence is not high, it tries an adaptive black and white version that removes shadows and uneven lighting, then the original colors. The best pass wins.

Up to a point. Capital letters about 20 to 40 pixels tall read best. Scanly enlarges smaller text automatically, but it cannot recover detail lost to blur or heavy compression.

Auto works for most pages. Choose “Photo, sign or label” when text is scattered, “Receipt or single column” for narrow slips, and “Single line” or “Single word” for tiny crops.

Full name, job title, company, email, mobile, work phone, fax, website, address and any leftover details such as social handles.

It scores each line on capitalization, length, position, whether it matches the email address, and whether it contains job or company words. The highest-scoring line wins.

Yes. Labels such as Mobile, Cell, Tel, Office and Fax are used to sort numbers into the right field. Extra numbers go into Other details.

It is a standard vCard that iPhone, Android, Outlook and Google Contacts can import, with name, title, company, phones, email, website and address in the right fields.

Use “CSV for CRM” for each card. The columns are the same every time, so the rows can be pasted into one spreadsheet and imported together.

Scan each side separately. One clear side per photo gives the most reliable result.

Vendor, date, time, receipt or invoice number, currency, subtotal, tax, tip, discount, total, payment method with card ending, and each line item with quantity and price.

It ranks lines such as “Grand total”, “Amount due” and “Total”, ignores “Subtotal”, “Total tax” and “Total items”, and falls back to the largest amount if no label is found.

It checks the arithmetic. If the items add up to the subtotal, or subtotal plus tax and tip equals the total, you see a green check. If not, it tells you what does not add up.

Yes. It recognizes symbols and codes such as $, €, £, ₹, ¥, Rs, PKR, AED and SAR, and both 1,234.56 and 1.234,56 styles. Amounts are not converted between currencies.

Faded thermal paper is hard for any OCR. The scanner uses an adaptive black and white pass first, which helps a lot, but a fresh, well-lit photo always gives the best result.

Yes. Download CSV includes the summary fields and every line item. JSON is available for developers and automations.

It first reads the real text stored inside the PDF, which is instant and exact. Only pages without a text layer, such as scans, are sent through OCR.

Yes. Scanned pages are detected and read with OCR automatically, using the language you choose. You can also force OCR on every page.

Yes. Enter a range such as 1-3, 5, 9- in the Pages box.

Yes. You are asked for the password, and it is used only on your device to open the file.

Yes. Line and paragraph breaks are rebuilt from the position of the text on the page. Tick “Join lines into paragraphs” for flowing text.

Text-layer pages are fast even in long documents. OCR takes a few seconds per scanned page, so very long scans take a while.

Yes. It keeps searching after the first code, so a page with several QR codes returns all of them.

In browsers with a built-in barcode detector, such as Chrome on Android, macOS and ChromeOS, it also reads EAN, UPC, Code 128, Code 39, ITF, Data Matrix, Aztec and PDF417. QR codes work in every browser.

Yes. Choose “Scan with camera” and point it at the code. The video is processed on your device and never recorded.

Links, Wi-Fi networks, contacts (vCard and MECARD), calendar events, email, phone, SMS, map locations and payment links, each shown with its fields and the right actions.

No. You see the full address first and choose whether to open it. Links are opened safely in a new tab.

The code may be too small, blurred or cut off. Crop closer to the code, avoid glare and make sure all four corners are visible.

Drag a row by its handle, or use the arrow buttons. Each image becomes one page in that order.

Yes. Each row has a rotate button. Phone photos are also turned the right way up automatically.

A4, US Letter, US Legal, A5, or “Fit each page to its image”. Orientation can follow each image or be fixed to portrait or landscape.

Images are saved as JPEG at the quality you choose. High keeps nearly all detail, Small makes a lighter file for email.

No. The PDF contains only your images.

The first run downloads the OCR engine and your language data, usually 2 to 15 MB. Later scans start almost instantly because the engine stays loaded and the files are cached.

There is no hard limit. Very large photos are scaled to a size that keeps text sharp without running out of memory.

Large images in Auto mode can take several passes. Press Cancel, then crop to the text or choose a specific enhancement under Advanced settings for a faster single pass.

OCR uses your processor for a few seconds per page, similar to editing a photo. Batches of many pages use more.

After the engine and your language have been downloaded once, scans usually keep working from your browser's cache, but this depends on your browser's storage settings.