Scanned PDF OCR

Add a searchable text layer to a scanned PDF, so its words can be selected, copied and searched.

PDF documents Heavy client-side (WASM) pdf.js rendering + tesseract.js OCR + pdf-lib text layer

Tool preview

How to use it

! The OCR engine (tesseract.js) plus the English language data (~15 MB) is downloaded from a public CDN on first use.
! Recognition is English only and capped at 20 pages, because it runs on your CPU.
! The page image is kept as it was; only an invisible text layer is added. Accuracy depends on scan quality.

Does the page look different afterwards?

No. The original image stays as the visible page, and the recognised words are drawn transparently on top at their detected positions.

Why is only part of my text selectable?

OCR can miss faint, rotated or stylised text. A sharper scan usually fixes it.

Is the PDF uploaded?

No, rendering and recognition happen in this tab.

Processing happens in your browser using the File API, WebAssembly and Web Workers. There is no upload endpoint.

Inspect a PDF author, producer, creation date and XMP data.

Delete the author, producer, timestamps and XMP from a PDF.

Permanently delete the text instead of covering it with a black box.