Copy text from any image. Paste a screenshot or drop a JPG, PNG, photo, scan or PDF and get text you can edit, check and export — tables, receipts and code included. Free online OCR that runs in your browser, so your images stay on your device.
Read on your device. Your images aren't uploaded. How this works
No image to hand? Try a sample:
Detection
Every image is checked for tables, receipts, code, documents and handwriting, and read in the way that suits it. You'll see a short label like "Looks like a table". Switch to another mode at any time without uploading again.
Looks like a table with 5 columns
| Region | Manager | Orders | Revenue | Growth |
|---|---|---|---|---|
| North | Priya Raman | 1,284 | $96,300 | 12.4% |
| South | Daniel Okafor | 967 | $71,850 | 8.1% |
| East | Mei Lin | 1,502 | $118,400 | 15.7% |
| West | Lucas Moreau | 1,118 | $84,920 | -2.3% |
Edit any cell, then copy it straight into Excel or Google Sheets, or download CSV or .xlsx. Numbers stay numbers.
Looks like a receipt
Four items add up to 29.50, matching the subtotal.
This photo was taken at an angle on a dark table. It was straightened by 2.2° and the lighting evened out before reading.
Looks like a screenshot of Python code
def invoice_total(items, tax_rate=0.055):
# items: list of (quantity, unit_price) pairs
subtotal = sum(Decimal(q) * Decimal(p) for q, p in items)
if subtotal <= 0:
raise ValueError("invoice has no billable items")Indentation is rebuilt from where each word sits, curly quotes are straightened, and the dark theme is inverted before reading.
Looks like an article or document
Caring for a Sourdough Starter
A sourdough starter is a living culture of wild yeast and bacteria. With a few minutes of attention each day…
Daily feeding
Wrapped lines become paragraphs, and headings and lists come through — including bullets the engine can't see. Download as Word or Markdown.
Read as handwriting
Meeting notes, Tuesday
Remember: water the plants!
Neat, print-style writing reads well. Joined-up cursive is harder, so doubtful words are underlined for you to check.
Found Devanagari text — read as Hindi + English
Welcome to the community library
सामुदायिक पुस्तकालय में आपका स्वागत है
Opening hours: 9 am to 7 pm
खुलने का समय: सुबह 9 बजे से शाम 7 बजे तक
Nothing to set up: lines in another script are spotted and the page is read again with the right language. 47 languages, several at once.
Keyboard
Most text people need is on a screen: an error message, a chat, a slide in a video call. Copy the screenshot, paste it here, and the text is ready a moment later. On a phone, share or upload the screenshot instead.
Review and export
Click a line of text and its spot lights up on the image; click the image and the cursor jumps to that text. Words the engine was unsure about are underlined, and one button walks you through them. You also get an honest confidence level, never a promise of perfection.
Drop a stack of screenshots or scanned pages. Each becomes a page you can reorder, read again or remove. The whole-document view joins them in your order, searches across every page and exports one Word, PDF or Markdown file — or a ZIP of separate files.
Copy keeps headings, lists and tables when you paste into Google Docs, Word or Sheets. Download any of eight formats, including a searchable PDF that puts selectable text behind the original image. Share a link that carries the text itself, so nothing is uploaded.
Privacy
The text engine runs inside your browser, so pictures of receipts, contracts, medical letters and private chats never reach a server. There's no account, no ads and no captcha.
Read exactly what happensSpecification
About the tool
How the image to text converter works, what it can read, and how to get clean, accurate text from any picture.
An image to text converter takes a picture that contains writing — a screenshot, a phone photo, a scanned page or a PDF — and turns that writing into real text you can select, edit, search and paste. The technology behind it is OCR, short for optical character recognition. Image to Text App uses Tesseract, an open-source OCR engine built on a neural network, and runs it directly in your browser. It doesn't just return a block of characters: it works out whether you've given it a table, a receipt, code, an article or handwritten notes, and lays the text out to match.
There's no sign-up, no daily cap, no watermark and no captcha. Because the text is recognised on your own device rather than on a paid server, the whole tool is free to use as often as you like, with every reading mode, every language and every download format included. Convert one quick screenshot or a stack of fifty scanned pages; nothing is held back for a paid plan.
It reads JPG and JPEG, PNG, WebP, GIF, BMP, TIFF and iPhone HEIC photos, up to 25 MB each. In a PDF, digital pages are copied exactly as they are and scanned pages are read with OCR, so turning a PDF image to text is quick and accurate. Screenshots of chats, error messages and slides read especially cleanly, and dark-mode captures are handled automatically. Neat, print-style handwriting works well too; joined-up cursive is harder for any on-device engine, so doubtful words are flagged for you. Each format has its own page with tips: JPG to text, PNG to text, PDF to text, screenshot to text and handwriting to text.
Copying keeps headings, lists and tables, so the text pastes neatly into Google Docs, Microsoft Word or Google Sheets. To convert an image to text in Word, download a .docx with its paragraphs and headings intact. Pictures of tables become an editable grid: fix any cell, then paste it into Excel or download XLSX or CSV with numbers kept as numbers. You can also save a searchable PDF, Markdown, JSON or HTML. See image to Word, image to Excel and image to PDF.
The converter reads 47 languages, from English, Spanish, French, German and Portuguese to Arabic, Chinese, Japanese, Korean and Russian. It also covers the major South Asian scripts: Hindi, Bengali (Bangla), Urdu, Nepali, Marathi, Tamil, Telugu, Kannada, Malayalam, Gujarati and Punjabi. Leave it on Auto and it spots lines in another script and reads them again with the right language, or choose several languages yourself for a mixed-language image.
Add up to 50 images or PDF pages in one go. Each becomes a page with its own result that you can reorder, read again or remove. The whole-document view joins every page in your order, searches across all of them, and exports one combined file or a ZIP of separate files — handy for a batch of receipts, lecture slides or a scanned report.
Many online OCR tools upload your pictures to a server. This one doesn't: the text engine runs inside your browser, so receipts, contracts, ID documents and private chats never leave your device. Only the engine and language files are downloaded the first time, and after that it keeps working offline. History stays off unless you switch it on. Read exactly what happens to your images.
Start with the sharpest image you have: a straight-on photo in even light, or a screenshot rather than a photo of a screen. Crop away anything you don't need, and pick the language if Auto isn't sure. For difficult images, the Adjust panel can straighten, sharpen and boost contrast before reading. There's more in the guide to getting accurate OCR results.
Paste a screenshot with Ctrl+V (⌘V on a Mac), or drag in a JPG, PNG, photo or PDF. Reading starts by itself and takes a few seconds per page. Check any underlined words, then copy the text or download it as Word, Excel, PDF or plain text.
Add the image and the text is extracted automatically. The tool works out whether it's a document, table, receipt, code or handwriting and lays the text out to match. To extract text from just one part of the image, crop it first; click any line to see exactly where it came from.
Paste or drop the image, wait a few seconds, then press Copy to put all the text on your clipboard, or select just the lines you need. Headings, lists and tables stay intact when you paste into Google Docs, Word or Google Sheets.
Yes, completely. There's no sign-up, no daily limit, no watermark and no paid plan: every reading mode, all 47 languages and every download format are included. Because the text is read on your own device, it costs almost nothing to run.
Yes. The text is recognised by Tesseract, an open-source OCR engine built on a neural network, and it runs in your browser rather than on a paid AI server. That means no credits, no free trial that runs out and no limit on how often you use it.
No. The text engine runs inside your browser, so your images never leave your device. Only the engine and language files are downloaded the first time and then cached, after which the tool keeps working offline. History stays off unless you switch it on.
Clear printed text, such as screenshots and scanned documents, is usually read very accurately. Small text, blurry or dim photos, decorative fonts and handwriting are harder, so words the engine is unsure about are underlined, and clicking a line highlights its spot on the image so checking takes seconds.
Yes. Neat, print-style handwriting usually converts well. Joined-up cursive is much harder for any on-device engine, so expect to correct some words; doubtful ones are underlined for you. A sharp, straight-on photo in good light makes the biggest difference.
Yes. Add the image, then download a .docx file with headings, paragraphs and lists kept, ready to edit in Microsoft Word. You can also copy the text and paste it into Word or Google Docs with its formatting intact, or combine several images into one Word document.
Yes. Drop in a PDF and every page is read, up to 50 at a time. Pages that already contain digital text are copied exactly, and scanned or image-only pages are read with OCR. Copy the result, or download it as plain text, Word or a searchable PDF.
Yes. Add a scanned or image-based PDF and every page is read with OCR. Then download a single .docx with the pages in order and headings, paragraphs and lists preserved, ready to edit in Word or Google Docs.
Yes. Tables are detected automatically and shown as an editable grid. Fix any cell, then paste it straight into Excel or Google Sheets, or download .xlsx or CSV with numbers kept as numbers.
Yes, up to 50 images or PDF pages in one go. Each becomes a page you can reorder, read again or remove. The whole-document view joins them in your order, searches across every page and exports one combined file or a ZIP of separate files.
47 languages, including English, Spanish, French, German, Portuguese, Arabic, Chinese, Japanese, Korean, Russian, Hindi, Bengali (Bangla), Urdu, Nepali, Tamil and Kannada. Leave it on Auto to spot other scripts automatically, or choose several languages for a mixed-language image.
Yes, on iPhone and Android in any modern browser, with no app to install. Take a photo with the camera button or pick a screenshot from your gallery, then copy, download or share the text. iPhone HEIC photos open without converting.
Yes. There's no account, no email address and no captcha. Open the page, paste or drop an image, and the text is ready in seconds. Nothing is saved unless you switch on History, and then it stays in your own browser.
Yes. JPG, JPEG and PNG all work, along with WebP, GIF, BMP, TIFF and iPhone HEIC photos, up to 25 MB each. Drag the file in, choose it from your device or paste it, and the text is extracted automatically.
Yes, and screenshots usually give the cleanest results. Copy a screenshot to your clipboard and press Ctrl+V (⌘V on a Mac) anywhere on the page; there's no need to save a file first. Dark-mode screenshots are inverted automatically, and code keeps its indentation.
Not with this tool. Image to Text App reads the text out of an image; it doesn't erase it from the picture. To remove text, use a photo editor with an object-removal, healing or magic-eraser tool, many of which are free online.
If you mean taking the text out of an image so you can edit or reuse it, yes: that's exactly what this tool does, and you can copy or download the result. If you mean erasing the text so the picture looks clean, you'll need an image editor with an inpainting or object-removal tool instead.