Bild in Text umwandeln

Paste, drop, or pick

Press Ctrl+V with a screenshot on your clipboard, drag images in, or click to select. Several at once is fine.

Set the language

Spanish is the default. Sixteen languages available, each loaded on demand. Getting this right matters more than any other setting.

Copy or download

Read the text, check the words flagged as low confidence, then copy it or save it as a .txt file.

A WebAssembly build of Tesseract does the reading locally. The only network requests are the one-time engine and language model downloads. Your image is never uploaded, so there is no copy of your document on anyone's server to leak, subpoena, or forget to delete.

The real use case for most people is a screenshot they took thirty seconds ago. Paste it straight onto the page. No temp file, no file picker, no upload dialog. Most OCR sites still make you save the image first.

Spanish, English, French, German, Portuguese, Italian, Catalan, Dutch, Polish, Turkish, Russian, Arabic, Hindi, Japanese, Korean, Simplified Chinese. Each model downloads once and is cached, so the tool stays fast without shipping 100 languages you will never use.

Every word carries a score. The summary counts how many landed high, medium and low, the shaky ones are listed with their percentage, and you can highlight them directly on the image. You proofread the five words that need it, not the whole page.

Grayscale plus a percentile contrast stretch by default, with an optional Otsu threshold pass. Tesseract was trained on clean high-contrast scans; giving it something closer to that is the cheapest accuracy lift available client-side.

Recognition reports progress as it works: loading the language, then reading the page. A large photo is scaled to 3000 pixels on its longest side first, because above that the engine slows down super-linearly with no accuracy gain.

Drop twenty images. Each gets its own text panel, confidence summary and download. One copy-all button and one combined .txt when you want the lot.

No account, no email, no daily quota, no page limit, nothing stamped on the output. There is nothing to meter, because there is no server doing the work.

Most free OCR sites upload your file, run recognition on their infrastructure and keep it for some period their terms describe vaguely. That matters when the image is an ID, a contract, a payslip or a medical result. Here you can open the network tab and confirm the image is not in any request.

The usual pattern is three free pages, then an email address, then a card. This tool has no account, so there is no wall to hit. Nothing about the design depends on converting you into a paying user.

Every OCR engine makes mistakes. Most tools hand you a block of text and let you discover them later, in the document you already sent. This one tells you which words it was unsure about, and shows you where they are on the image.

Spanish is the default language, not an afterthought behind English. Accented characters and the inverted punctuation come through correctly because the right model is loaded from the start.

Copy text out of a screenshot

An error message, a chat you cannot select, a slide in a video call. Screenshot it, press Ctrl+V here, get the text back in seconds.

Digitise a receipt or invoice

Photograph the paper, set preprocessing to Binarize for a dim shot, and pull the line items out as text you can paste into a spreadsheet.

Get quotes out of a book page

Photograph the page, read it in the right language, and copy the passage instead of retyping it. Check the flagged words before you cite it.

Turn a scanned document into editable text

Drop the scan, copy the output into your editor, keep the formatting decisions to yourself. Nothing about the document reaches a third party.

Extract text from an infographic or chart

Labels, legends and axis text that were never selectable. Batch the whole set and download one combined .txt.

Read a foreign-language sign or menu

Photograph it, pick the language, get the characters as text you can paste into a translator. Two steps, no account for either.

Convert a screenshot to text in five seconds

1. 1.Take your screenshot so it is on the clipboard.
2. 2.Come to this page and press Ctrl+V (Cmd+V on Mac). No need to click anything first.
3. 3.The language model downloads once, then recognition starts. Watch the progress bar.
4. 4.Click Copy. The text is on your clipboard.

Get better results from a phone photo

1. 1.Shoot the page flat and filling the frame. Perspective and curl hurt more than resolution does.
2. 2.Drop the photo in and set the language to match the text.
3. 3.If the output looks noisy, switch preprocessing from Grayscale to Binarize. That re-runs recognition automatically.
4. 4.Check the low-confidence word list before trusting the result.

Convert a batch and get one text file

1. 1.Select or drag every image at once. They queue up and run in order.
2. 2.Each file gets its own panel with its own confidence summary.
3. 3.When two or more are done, a batch bar appears at the top.
4. 4.Click Download all to get a single .txt with every result, separated by blank lines.

Find and fix the words it got wrong

1. 1.After a conversion, look at the three confidence chips: high, medium, low.
2. 2.The words below the medium threshold are listed underneath with their scores, worst first.
3. 3.Click Show uncertain regions to draw those words as boxes on the image itself.
4. 4.Fix them in the text box, then copy or download.

Ctrl+V and the text is yours. The fastest path from an unselectable screen to a paste-able string.

Receipts, invoices, forms, book pages. Binarize preprocessing plus the confidence list turns a phone photo into something you can trust enough to paste.

If you would not email the document to a stranger, do not drop it on a site that uploads it. This one has nothing to upload it to.

Spanish first, plus 15 more languages including Arabic, Russian, Japanese and Chinese. Each model loaded only when you ask for it.