Job ticket · PDF bench
A scanned PDF is a photograph of paper: you can look at it, but you cannot search it or copy from it. This page reads the words off the pages, shows you the text to check, and hands it back as text and as a searchable PDF. Everything runs in your browser; the file is never uploaded.
Drop a scanned PDF on the tray, or click to pick one
PDF · it stays on this device
Each page is drawn at 300 DPI and read. One page takes a second or two once the engine is loaded; the first run loads it (about 10 MB, once). The text lands in an editable box below: it is yours to check and fix before anything is downloaded.
Pick a PDF to begin.
The sharpness matters more than anything: on our bench the same page read at 200 DPI fell apart (about 62% of characters right), while at 300 DPI it read 96.1%. So this page always renders at 300 DPI before reading, even when the scan inside the PDF is smaller. The recogniser also reports where each word sits, which is how the searchable PDF gets its invisible text layer.
Clean, straight scans read well; skewed phone photos, shadows, handwriting and stamps over text will read worse, sometimes much worse. Numbers are the weak spot: this engine dropped a decimal point in a money amount on a clean test page, so check figures character by character. Files up to 50 pages read in one go. English is the language loaded in this version.
The page draws each page of your PDF as a picture at 300 DPI, the sharpness the reading engine needs, and the engine reads the words off the picture. On our test invoice, a clean one-page scan, it read 96.1% of the characters correctly.
Check them. On our clean test invoice this engine read the total 124.50 as 12450, dropping the decimal point. That is why the text is shown in an editable box: read it, fix anything, then download. Never send on numbers you have not eyeballed.
The text in the box (downloadable as a .txt, with your fixes included) and a searchable PDF: your pages as pictures with the recognised words laid invisibly over them, so the file can be searched and copied. The searchable PDF carries the text exactly as read, before your fixes.
Nowhere. Scans hold some of the most private paper people own, and this page was built for that: the file is read into this page and processed in your browser. Watch the network panel while it works if you want to check.