Skip to content
naina

Extract text from any document.

PDFs and images, in your browser, free and without limits. Your file never leaves this tab. There is no server to send it to.

Drop a document

PDF or image, paste, or click to choose

Options
Nothing leaves this tab. There is no server. The reading runs here, in your browser. Once the models are cached there is no network traffic at all.
Pick the right script. A wrong script gives wrong text that looks right, not an error. Auto detects it for you. Handwriting is unreliable. Full limitations.

How this works

Nothing is uploaded

Everything runs in this tab, PDF rendering included. No account, no quota, no upload.

Works offline

The models are cached on first use. It reads with no connection.

The same code as the server

The browser runs the same C++ core as the Python and Node packages, compiled to WebAssembly.

What it cannot do

Printed Latin, Chinese, Japanese and Korean read well. Handwriting is weak. Scripts it was not trained on, Devanagari among them, come back as confident nonsense.