Inside the JAWS 2018 betas: smarter, more powerful OCR

This post highlights a new JAWS 2018 feature we’ve been getting some terrific feedback about, our expanded optical character recognition (OCR) capabilities.

For some time, JAWS has offered a feature which has helped many of us out immensely, the ability to perform optical character recognition on certain types of graphical images. Most of us have experienced the challenge at least once. Someone sends an important document such as an invoice as a PDF file. We open the document to find out how much we owe, only to discover that the PDF file contains the image of a document, not its actual text. In a situation like that, it’s tempting to conclude that if the document is blank to us, then we must owe nothing, but we know we’ll be caught up with in the end.

Unless a document contains something unusual such as a picture of a hand-written document, I’ve had excellent results opening graphical PDF files in Adobe Reader, then pressing the JAWS Key+Space, O for OCR, and D for document. JAWS recognizes text in the file, and displays it in the Results Viewer with the virtual cursor active, just as if it were on a web page.

Customers tell us that they continue to advocate for those they do business with to provide them with content that is truly accessible, but when you simply must get at the text of something, this JAWS feature can make the difference between being able to perform one’s job and not. That’s what we’re all about.

Even before JAWS 2018, the convenient OCR features have gone beyond PDF files. Customers use the JAWS OCR technology to read data that has been embedded graphically rather than textually on a web page. Not a good thing for web designers to do, but when the job must be done, JAWS delivers.

In JAWS 2018, we’ve expanded the OCR smarts even further. In addition to extracting text from PDF images, JAWS 2018 adds support for BMP, JPG, JPEG, GIF, TIF, TIFF, PNG and PCX files.

There are two ways to use this feature. You’re likely to use the first one if you only require OCR occasionally, because it’s menu-driven and easy to remember. If you find yourself in need of the recognition feature a lot, it’s worth remembering the keystroke, to maximize your efficiency.

If someone sends you a graphical file that you know contains text, save it to your computer, locate it in File Explorer, then press the Application key or shift+F10. Right on the context menu, you’ll find an option called “recognize with JAWS”. Press Enter, and the text recognition begins.

Alternatively, when focussed on the file, press the JAWS Key+Space, then O (for OCR) then F (for file).

The OCR improvements don’t stop there. As a blind person myself, I’m grateful to be dealing with much less print than I was 20 years ago. We might not quite be a paperless world yet, but there’s less paper in the average office than there used to be, and that’s great news from an accessibility point of view.

However, there are still times when it’s useful to snap a picture of a print page and find out what it contains. Now you can acquire an image right from within JAWS, in conjunction with the Freedom Scientific Pearl® camera, or a flat-bed scanner. If you use our OpenBook scanning and reading solution, you’ll be familiar with the terminology we use when reading a paper document, “acquire a page”. We use the word “acquire” because you might be accessing the printed page either by taking a picture of it, or scanning it, depending on the hardware you have. So, with a compatible device connected and a page positioned appropriately, press the JAWS Key+Space, O for OCR, and A for acquire.

With the portable and capable Pearl camera, it’s easy to take pictures of multiple documents or multipage documents, using Pearl’s motion detection.

You can learn much more about the new OCR capabilities of JAWS 2018 from the JAWS what’s new material on our website, and hear a demonstration in FSCast 143.

One of the most enjoyable parts of the public beta process for us is when we roll out a feature we’ve been working on for a while and get feedback that it’s a hit. Just to give you one example, Jonathan Nixey from the UK was so impressed by what he was experiencing in JAWS 2018, he just had to drop us an email to share his enthusiasm. He writes in part,

I just had to share this with you all.

The improved optical recognition feature in JAWS 2018 had me curious as to how it may cope with very old newspaper articles. As someone who is very keen on family history, old newspapers can hold a great deal of information.

Just the other day, someone emailed me a jpg scan of a newspaper article I was interested in. It dates from April 1866, so is over 150 years old, and is about our local Railway Station Master, William Mogford. I wasn’t expecting miracles, but JAWS produced a miracle!

In fact, I sent the scanned text that JAWS 2018 had recognised to a professional genealogist friend in London who has transcribed numerous newspaper articles for me over the years. I simply had to share her comments with you! This is what Judy had to say:

“I think this is a stunning piece of reading by JAWS. Really there are hardly any errors — just Mogford, a few stray pieces of punctuation, and a handful of misspellings. Part of the type is quite blotchy towards the end, and frankly I’m surprised JAWS did so well.”

I just felt the need to share with you Judy’s very complementary thoughts on the ability of JAWS 2018.

Keep up the good work.

Whether you’re performing OCR on a desktop in the office, a laptop or tablet, or of course an ElBraille which gives you note taking without the compromises inherent in a mobile operating system, we hope you’ll enjoy all the new OCR capabilities in JAWS 2018.