So I just know that they exist, and they are older probably than NVAccess, same legacy ish company as Freedom Scientific.
Why the heck would anyone want to pay +$1000 for an OCR software in 2025 especially considering that Pneuma Solutions offering is so so much better, and actually maintained and upgraded? Okay the only downside is that as far as I know last time I tried Scribe it's technically a bit inconvenient to actually scan from the web with a flatbed scanner, from Double Tap I know that they are actively working to patch the only real pseudo arguments some people have for the other legacy style softwares. Is Kurzweil as bad as OpenBook in terms of unpatched bugs for modern windows versions, which I unfortunately happen to have through RAMQ, my only real regret of government money wasted in recent years.
By TheBlindGuy07, 29 August, 2025
Forum
Assistive Technology
Comments
All depends on what you want to do.
If you want to scan a book, or many books, you need something robust. I doubt if any newer software meets the criteria.
Kurzweil 3000
Back in my early college days, I used Kurzweil 3000 for the iPad. I will say that, for what it does, it is very good. That is to say, use it to read textbooks. I would not dare use it for casual or leisurely reading though.
My personal issue with this type of software is that instead of just using gestures to read, like you would on Kindle for example, you have buttons for reading forward, going backward, reading by character, reading by word, reading by line, etc. Everything is its own control, which makes it a tedious implementation of software use.
Just my two cents.
Abandonware
K1000 hasn't been updated to a new version since 2013 or 2014. They've put out some maintenance patches. I still use my copy from 2012 or 2013, because, like Chris said, the simple/free stuff doesn't cut it. Not just for books, but for semantics. ABBYY Fine Reader has always been great at parsing sentences and the like to actually infer accurate OCR results. ScanSoft wasn't at all far behind. Jaws uses an older ScanSoft engine for its Convenient OCR, so it's pretty good. ABBYY now is an expensive subscription model, but it uses machine-learning. I'm pretty sure a paid A.I. subscription can get you document recognition by now, but I haven't looked into it. With a subscription to a Vision model like GPT4V or Gemini Vision Pro, you can send images page by page and ask for the text. Or so Copilot tells me. Copilot says a lot of nonsense, though.
Kurzweil 1000
Hi,
I had Kurzweil 1000 years ago for scanning books. I had to scan a 900 page book for work once. It was my last resort, because there were no online versions available for that particular up-to-date edition. So I got to like 800 pages, and I happened to check on the off-chance whether it was now available on Kindle, and it was. So all that work for nothing. I tried scanning another couple of language learning books, but I just didn't have the heart to finish them. I had a plustek flatbed scanner, which was too big considering I hardly ever used it in the end, so I got rid of it. If I need a book, it's usually available online these days anyway. I wouldn't go back to Kurzweil and a scanner. And yes, you can send images of scanned PDF pages to an AI. I've done that with ChatGPT plus and the paid version of Gemini I'm on, and the AI Content Describer add-on with NVDA, and the results are excellent. There have only been a few times it's halucinated.
This was life-saver in old days
I was using this almost daily in my college as professors were constantly giving new handouts and reading materials or notes in printed form. Sometimes, there would be a range of pages to read, so, I would photo-copy the book pages and feed those coppied pages in a scanner with auto-feeder. Scanning 1 page at a time from the original intact book manually would be so time consuming.
Kurzweil1000 on a windows PC did a decent job of recognizing paragraphs and tables, however it absolutely failed when it came to diagrams and charts. Still, it was my default option as at least reading paragraphs around those charts and graphs gave me a fair idea of what was being discussed.
Back in those days KNFB reader was super popular on phones, before Seeing AI dethrowned it. Seeing AI actually does an impressive job of table recognition. It's still my default scanner app on phones despight the all AI apps.
I don't know about any updates to Windows app either, there was something called as Kurzweil3000, not sure if that was it's updated version.
I wish if Microsoft brought Seeing AI app to Windows, with it's decent table recognition and all, it would be a solid solution.
As of now, I relye on NVDA Advance OCR Addon and Bookworm's OCR capabilities. I hope Bookworm doesn't get abandon, hasn't seen an update to it over the last year. Both of these however, fail to deal with tables.
I also sometimes just open the PDF files directly in Microsoft word, and most of times, it gives me an accessible file to read even if the PDF was just scanned pages. And mine is not even Office 365, it's just the free perpetual office 2019 license that my laptop shipped with. It does a decent job, and even handles tables if I remember correctly.