3 Replies Latest reply on Dec 4, 2017 6:24 AM by laynie724

    Recognize Text: Not working on large files.


      I need to be able to send attorneys PDF files that contain thousands of pages of PDFs. (It needs to be in one PDF.) When I use the Recognize Text feature and run it through the file it appears to be converting since I see the status "Converting scanned page to Searchable Image", but when it's complete and I go to do a search (Control + F) it doesn't work.


      Has anyone else experienced this and know what to do to fix this? Or a workaround. This is a functionality that we need to have.



        • 1. Re: Recognize Text: Not working on large files.
          nickb96703691 Level 1

          I've noticed that long docs (400 + pp) take a lot of quiet time at the end to consolidate fonts etc. and your PC might go to sleep before finishing.  That and/or other interruptions can KO the whole process.  I'm not an expert but I'd try splitting super-long docs into 100 page sections, then optimizing or OCRing each section, and then joining them back together.  That way you lose less time if one section goes south and needs to be done over. 

          1 person found this helpful
          • 2. Re: Recognize Text: Not working on large files.
            Test Screen Name Most Valuable Participant

            It might also be that the original is not suitable for converting to text. For example, handwritten material won't convert. Poor scans, scans that aren't written as English paragraphs with correct spelling, weird fonts: all these have low success rates. It's all guesswork, not an exact science.

            2 people found this helpful
            • 3. Re: Recognize Text: Not working on large files.
              laynie724 Level 1

              Thanks for the reply! In this particular instance where I found this to be a problem was where the PDF file had over 7,000 pages. The thought of breaking it into smaller chunks was our resolution, although tedious. I get that it's not exact science. Just wondering if there was any other workaround out there.