Image to Text (OCR)

1. 1Add an image containing text.
2. 2Choose the language of the text, which improves accuracy considerably.
3. 3Press Extract text and watch the progress — the first run downloads the language data.
4. 4Review the result, then copy it or download it as a .txt file.

This uses Tesseract, the long-established open-source OCR engine, compiled to WebAssembly and run inside a Web Worker in your browser. The engine finds text regions, segments them into lines and characters, and matches character shapes against a trained model for the language you selected.

Accuracy depends almost entirely on the input. A flat, well-lit screenshot of printed text is usually near-perfect. A photograph taken at an angle, in poor light, or of handwriting will be much less reliable — Tesseract is trained on printed type, not handwriting.

Choosing the right language matters more than people expect, because the model uses a dictionary to resolve ambiguous shapes. Running English text through the German model will produce noticeably worse output.

The language data is a few megabytes and is downloaded once, then cached for later runs. Recognition itself happens entirely on your device — the image is never sent anywhere.