Free Image to Text Converter
This free tool reads the printed text out of a screenshot, a scan, or a photo of a page and hands it back as text you can copy, running the recognition engine inside your own browser so the image is never uploaded.
Drop an image here
A screenshot, a scan, or a photo of a page
The first read downloads about 5 MB. That is the recognition engine itself, which then runs on your device instead of on a server — so the picture never leaves your browser. Your browser keeps the engine afterwards, and later reads start immediately.
Runs entirely in your browser. Your images are never uploaded to a server.
How it works
- Drop an image onto the box above, paste a screenshot straight from your clipboard, or click to choose a file.
- The first read downloads about 5 MB of recognition engine — by a distance the largest download on this site. It happens because the engine runs on your machine instead of on a server, and your browser keeps it, so later reads start immediately.
- The text appears in an editable box alongside a confidence score, so you can correct anything the engine misread while the picture is still on screen next to it.
- Copy the text with one button, or download it as a plain .txt file.
Common use cases
- Getting the words out of a screenshot — an error message, a chat, a slide you cannot select text on.
- Turning a photo of a printed page, receipt, or letter into something you can search and paste.
- Lifting a reference number, address, or serial off a photographed document instead of retyping it.
- Reading a passport, payslip, or medical letter without handing the image to a stranger's server.
- Recovering the text from an old scan, or from a PDF page you have exported as an image.
Frequently asked questions
- Is my image uploaded to a server?
- No. What moves across the network is the recognition engine coming down to you, not your picture going up. Once it has loaded, the image is read by your own browser and the text never leaves the device. That is the reason this tool exists: almost every other free OCR site uploads the file, and the documents people run through OCR — passports, payslips, medical letters — are the last things worth uploading anywhere.
- Why does this tool download about 5 MB?
- Because the recognition engine is real software rather than a canvas trick. It is Tesseract, compiled to WebAssembly, plus a trained English language model, and both have to be on your device for the reading to happen on your device. Most tools here are a few kilobytes because resizing or cropping a picture is arithmetic; recognizing letters is not. The PDF tool is the only other one that loads a real engine, and this one is several times larger again. The alternative is sending your image to a server that already has the engine, which is exactly the trade this tool refuses. Your browser caches it, so it is a first-visit cost, not a per-image one.
- What kind of images does OCR read well, and what does it struggle with?
- It is good at printed characters: screenshots, PDF pages exported as images, flatbed scans, and photos of a page shot square on in decent light. It struggles with handwriting, script and heavily stylised lettering, text laid over a busy photo, anything smaller than roughly ten pixels tall, and pages photographed at an angle so the lines curve. The single biggest improvement is usually resolution — a closer, sharper photo beats any amount of correcting afterwards. The model here reads English.
- Can it read handwriting?
- Not reliably, and it is worth saying plainly rather than letting you find out. Tesseract is trained on printed type, so handwriting comes back as confident nonsense rather than as an obvious failure. Neat block capitals sometimes survive; ordinary cursive does not.
- What does the confidence percentage mean?
- It is the engine's own average certainty across the characters it read, and it is shown because OCR fails quietly. A bad photo does not produce obvious gibberish — it produces a fluent-looking sentence with a digit changed, which is far more dangerous. Above about 85 percent the mistakes are isolated characters. Between 70 and 85, whole words come back wrong while the text still reads normally, so check it against the picture. Below 70 the engine was guessing at shapes, and taking a better photo will help more than editing will.
- Can I use it on a PDF?
- Not directly — this tool reads images. Convert the pages to images first with the PDF to JPG tool on this site, then drop one here. That runs in your browser too, so a document still never leaves your device at any point in the chain.
Related tools
Understand what this does
Plain explanations of what is happening to your file, and what it costs you.