OCR PDF

How to make a scanned PDF searchable

1. Choose the scanned PDF, or photos of the pages (each photo becomes a page, in the order you chose them).
2. Choose the language: Telugu + English (for pages with both, or when you are not sure), Telugu or English.
3. All pages are read, or type the ones you want (1-3, 5).
4. Tap Make it searchable, then download the searchable PDF, or copy the text or save it as a .txt file.

The pages look exactly as before: the scan itself is not changed, not even redrawn. The words that were read lie over each page as invisible text, each word over its place, so you can search the PDF, select a word and copy it in your PDF app, such as Chrome, Adobe Acrobat Reader, Preview on a Mac or the Files app on an iPhone.

What happens to each page

• A page that already has text of its own (a PDF made on a computer, or one read before) is left as it is: it can already be searched. If its text copies as wrong letters, as old Telugu fonts often do, tick “Read the pages that already have text again”.
• Every other page is drawn at 300 dots an inch (a scan at its own sharpness), made grey, its contrast stretched so faint scans and yellowed paper read like clean ones, and its specks of dust taken out.
• Tesseract reads it: the open-source OCR engine, with its Telugu and English models, running in your browser.
• A page lying sideways or upside down is found and read the right way up; the result says which, and can turn them upright in the PDF too. (For a page scanned at a slight slant, use Straighten Scanned PDF first.)
• The result shows how sure the reading was on each page, and lists the words it was least sure of, so you can check those against the page.

Better Telugu reading (a switch, on by default on computers) has a second Telugu reader look again at the lines Tesseract was unsure of, and keeps the better word of the two, using Telugu Typing’s word list. On our hand-checked pages it took the Telugu words read wrong or missed from 57 in 749 to 25; it makes reading up to three times as long and needs up to about 250 MB more memory, so it starts off on phones. It downloads about 22 MB once, from ap7am, and is kept with the models; if it cannot run (an older browser, too little memory), the text is read with Tesseract alone and the page says so.

Everything happens in your browser: the PDF is never uploaded to ap7am or anywhere else. The first time, your browser downloads the reader and its models from ap7am, about 6 MB for Telugu + English (about 3 MB for Telugu alone), and keeps them, so the next time it starts at once. “Forget the downloaded models” lets them go, Better Telugu reading’s too.

What it cannot do

• OCR makes mistakes, more often in Telugu (letters with a vottu, such as ్య and ్వ, look alike) and in small or faint print. Check names, numbers and dates against the page before you rely on them, for example in a form.
• Handwriting is not read, nor text in photos of signs or in decorative lettering.
• Tables come out as lines of text, not as a table: to take tables out of a PDF that has text, use PDF to Excel.
• A locked PDF can be read with its password, but its pages go in as pictures and the new PDF has no password. To keep the pages exactly as they are, unlock it first with Unlock PDF.
• Reading takes time: a page full of text takes a few seconds on a computer and 10 to 20 seconds on a phone.

Questions

Made with Tesseract OCR (tesseract.js and tesseract.js-core, Apache 2.0 licence) and its Telugu and English models (Apache 2.0, packaged under the MIT licence), harsha-desaraju’s Telugu OCR model (MIT licence) and ONNX Runtime Web (MIT licence) for Better Telugu reading, pdf-lib (MIT licence) and Mozilla's pdf.js (Apache 2.0 licence), all running in your browser. Licences.