Hack the HydraKey facts
| Input formats | Images (PNG, JPG, WebP and other browser-readable formats); 7 OCR languages |
|---|---|
| Size limit | 100 MB per file (checked in the browser) |
| Where it runs | In your browser with Tesseract.js 5.1.1; the file is not uploaded (the library itself is downloaded from a CDN) |
| Cost | Free, no sign-up |
| Output | Plain text with a confidence score, copy or download as .txt |
How it works
The image is read by Tesseract, an open-source OCR engine, running inside your browser tab. The first run downloads the language data, then recognition takes a few seconds per image. Line breaks follow the lines in the image, and the tool reports a confidence score: a clean screenshot usually scores above 90%, a blurry phone photo much lower.
Choosing the language
Pick the language of the text before you start: English, French, German, Spanish, Italian, Portuguese or Dutch. The language model knows which letters and accents to expect, so the wrong choice turns é into e or ß into B. For a mixed-language page, choose the main language and proofread the rest.
Formats and limits
PNG, JPG, WebP and most images your browser can open work, including screenshots pasted from the clipboard. Printed text is read well; handwriting, stylised logos and text on curved surfaces are not. The output is plain text: tables lose their columns and must be checked by hand. For a scanned PDF, use OCR PDF, which reads every page.
Example
A clean screenshot-style image with three lines of text, OCR language English. Line breaks are kept; the tool also reports a confidence score (95% here).
Input
(PNG image, 900 x 220 px) Invoice 2026-0142 Total due: 1,250.00 EUR Payment within 30 days.
Output
Invoice 2026-0142 Total due: 1,250.00 EUR Payment within 30 days.