Add your files
Drop one PNG or two hundred. Batch conversion is not a paid feature here.
Text is trapped in a screenshot, a scanned page or a photo of a slide, and retyping it is not an option.
This is not a format change, it is recognition. Tesseract in WebAssembly segments the image into lines and characters and outputs plain text. Layout, fonts, colour and images are discarded; only characters remain.
OCR output always needs proofreading, and it is worst exactly where it looks most confident: zero against capital O, one against lowercase l, and column boundaries. Our engine reports its confidence per run so you know how hard to check.
No. The recogniser, its WebAssembly core and the language data are all served from our own origin and run in your browser. It works offline once cached.
Roughly 300 DPI, or text about 20 pixels tall. Below that the accuracy falls off sharply.
No. Tesseract is trained on printed type. Handwriting needs a different class of model entirely.
Drop one PNG or two hundred. Batch conversion is not a paid feature here.
Plain Text is lossless, so nothing is thrown away in the conversion.
Instant, because the file never went anywhere. Nothing expires, since nothing was stored.