A scan is just an image: you can't search it for a word or copy a sentence. OCR recognizes the text on each page and adds it to the PDF without changing how it looks, or saves it to a plain text file.
How it works
- Select the scanned PDF.
- Choose the document language.
- Run OCR and download the searchable PDF or the text.
Frequently asked questions
Which languages are recognized?
French, English, Spanish, German, Italian, Portuguese, Dutch, Chinese, Japanese and Hindi. The “French + English” option suits documents that mix the two.
How long does it take?
A few seconds per page. The first time, the recognition engine is downloaded (a few MB).
Will my PDF look different?
No. With “Searchable PDF”, the pages look the same: the recognized text is added invisibly so you can select and search it.
Can I get just the text?
Yes, choose “Text file”: you get a .txt file with the recognized text, page by page.
Does OCR recognize handwriting?
It's designed for printed text. Very legible handwriting may be partly recognized, but the result isn't reliable.
Can I run OCR on a photo?
First turn the photo into a PDF with “Scan to PDF”, then run that PDF through “OCR PDF”.
Are my files sent to a server?
No. Everything happens in your browser: your documents never leave your device.
What OCR does
Optical character recognition (OCR) reads the text in the image of each page. PDFRoll gives you two output options:
- “Searchable PDF”: the PDF looks exactly the same, but an invisible text layer is placed under the image. You can then select, copy and search the text.
- “Text file”: just the recognized text, page by page, in a .txt file.
The “Skip pages that already contain text” option, checked by default, skips pages that already have text and saves time on mixed documents.
Choosing the document language
The “Document language” menu offers French on its own or combined with English, as well as English, Spanish, German, Italian, Portuguese, Dutch, Chinese, Japanese and Hindi. Choosing the right language greatly improves recognition, especially of accented characters.
The first time you use it, the recognition engine is downloaded (a few MB). After that, allow a few seconds per page, depending on how powerful your device is. The document itself stays on your device: your browser does the recognition.
Getting accurate recognition
- Start from a sharp, high-contrast scan: blurry or faint text is poorly recognized.
- Turn upside-down pages the right way up with “Rotate PDF” before running OCR.
- Printed text is recognized far better than handwriting.
- Double-check important numbers and proper names, since OCR can make mistakes.
If no text is found, check the selected language and the scan quality. For a long document, you can also process only the pages you need by isolating them first with “Extract pages”.
Before and after OCR
To turn photos of documents into a PDF, start with “Scan to PDF”. Once the text is recognized, you can bring it into an editable document with “PDF to Word”, or extract all the text with “PDF to text”.