Reading the text in a scanned PDF with Read by Zaweba

A scanned PDF looks like a document but behaves like a stack of photographs. You can see the words, yet you can't select, search or copy them, because nothing in the file says they are words. Read by Zaweba fixes that by opening the PDF, turning each page into a picture and reading the text from it.

It's one of the most common reasons people need text recognition at all. Banks, solicitors, schools and offices send scans as PDFs, and old paperwork scanned years ago is usually sitting in PDFs too.

How Read handles a PDF

Read renders each page with PDFium at 300 DPI. That resolution is a good match for text recognition: high enough to show small print clearly, without making pages so large that reading slows to a crawl. Each page then goes through the same steps as any picture: clean-up where it's needed, then reading, then review.

Some PDFs can't be opened. A password-protected file, or one that is damaged, gets a plain message saying so, rather than a crash or a blank result. If a file is protected, open it in the program you normally use with the password, save an unprotected copy if you're entitled to, and give Read that.

Checking a multi-page result

A scanned PDF often runs to many pages. In the review, a multi-page document has a page list with thumbnails, so you can step through it. The cleaned-up page and its text sit side by side. When Tesseract is the engine, words it is less than 70 per cent sure of are shaded amber in both places, and F8 jumps to the next one. Click a word on the picture to find it in the text, or the other way round.

The text is editable, and your corrections are used when you copy or save.

What to save it as

What you do next depends on what you need. If you want the same document, but searchable, save it as a searchable PDF. Each page stays the cleaned-up picture, with every word laid over it as invisible text, so Ctrl+F and copying work. Corrections you made in the review go into that text layer.

If you want to edit the words, save as a .docx. Read writes paragraphs, line breaks, tables, page breaks and two-column sections, and you don't need Word installed to do it. If you only need the words, save a .txt file or copy all the text. Tables can go out as CSV, on a best-effort basis.

Many PDFs at once

When there's a folder of scanned PDFs to get through, the Batch tab reads them in the background, either into one combined document or one set of files per input, in whichever formats are ticked on the Output tab. Jobs queue, show their progress and the number of words to check, and can be cancelled. Files that can't be read are reported without stopping the rest.

Getting a better result

The scan itself matters most. A crooked or faint page can be helped by Read's straightening and contrast boost, but a page scanned too small or too blurry has lost detail that no software can bring back. If you're scanning the paper yourself, Read can drive a WIA scanner directly at 150 to 600 DPI, which is often better than scanning to PDF first. Read by Zaweba does all of this on your own PC, so the scanned documents never leave it.