Loading your local toolbox…
A scanned PDF is a photograph of paper. You cannot search it, copy from it, or convert it to Word until something reads the letters. That is OCR.
No upload. No watermark. No sign-up. Every tool mentioned here runs inside your browser.
Each page is rendered at a chosen DPI, the Tesseract engine recognises words and their bounding boxes, and those words are written back into the PDF as invisible text positioned under the image.
The page still looks identical — the text is transparent. But now search, copy and screen readers work.
Scan quality dominates everything. 200–300 DPI, straight, evenly lit, high contrast. A crooked phone photo of a curled page will produce garbage regardless of the engine.
Deskew and crop first if the scan is skewed — the Crop and Resize tools help. Higher DPI also helps small print, at the cost of processing time.
Language matters: the bundled model is English. Text in other scripts needs the matching language pack.
Handwriting. Modern OCR is poor at cursive — expect partial results at best.
Tables. OCR reads text in reading order, not cell by cell, so a table becomes a run of words.
Mathematical notation, unusual symbols and decorative fonts.
No. The engine (WebAssembly) runs on your device. The first use downloads it once, then it works offline.
Slightly — the invisible text layer adds a little. The page image dominates the size either way.
Loading your local toolbox…