The magazine of the Melbourne PC User Group
Dragon Naturally Speaking
David Hague
 

"You talk, it types" used to be a marketing slogan for voice recognition software. That claim wasn't quite true. But voice recognition has come a long way since then, says David Hague.

Over the years I've kept a close eye on voice recognition programs. Maybe it's the Captain Jean-Luc Picard in me, but I've always wanted to say "Computer, ..." whatever I felt I wanted the computer to do at the time.

But after many hours training every conceivable voice recognition system on the market, I never really found one that worked. Or worked continuously anyway. They seemed to know whenever you had a cold, sore throat or even had a sip of a nice red, as under these circumstances, perfectly formed English came out as Klingon. As the TARDIS doesn't have voice recognition, maybe the Doctor has it right and it is an impossibility. I almost gave up.

But then along came Dragon Naturally Speaking version 9 promising that it worked straight out of the box. No training needed, just a quick few words to get the microphone levels right, and away you'd go. And paint me blue and call me a bus, it works!

Not only can I dictate almost flawlessly all of the stories and numerous emails I write, but unlike previous versions, DNS 9 works in seemingly every program both for entering data and controlling the program and Windows itself. For example, to load Word, create a document, save it, print it and then email it, the spoken sequence of commands would be:
Say "Wake up" to turn on the microphone; say "Start Microsoft Word" to load Microsoft Word.
Next you'd dictate your document complete with punctuation such as "full stop", "tab", "new paragraph" and so on.

You can also get DNS9 to "read back" your text by saying "Play that back".

Say "File Save" opens the file menu and then the save dialogue box. Say "File Print" opens the file menu and then the print dialogue.
 
Say "File Send" opens a dialogue to send the file as an attachment.

To fill-in fields, simply speak the words required. To move from field to field, you can say "Tab" and any other keys on the keyboard can be "pressed". You can also say "Press A" or "Press X" to "press" these keys.

DNS9 is remarkably easy to use and understand. There's a vast array of possibilities available (the manual is over a hundred pages), but most of what you want to do is common sense. For example, if you have a sentence that requires italicising, you move the cursor to the sentence to format by using keyboard, mouse or voice commands. Next, you say "select sentence" followed by "italicise that". Selection of text is very flexible; as equally as valid is "select next two words", "select next three characters" as well as "select paragraph" and so on. The same syntax is used for bold, underline and even adding bullet points.

For special characters such as TM ® simply say "trademark sign" or "copyright sign" and so on. For email and web addresses, the full wording is spelled out such as "editor at melbpc dot org".

Finally, although DNS9 does understand a plethora of names, locations and so on, if you come across something that there's no chance it will understand, for example, "Meekatharra", then it can be quickly trained on the fly so the next time around there won't be an error.

For those with up-market dictation machines with USB memory, you can use these to dictate first and then transcribe later. DNS9 will read in the audio file and transcribe it for you to text. How cool is that! And it also works with headsets with Bluetooth allowing you to whimsically wander around your office dictating your next literary masterpiece, or take your laptop with you into the wooded dell countryside by the babbling brook, to discover your inner poet.

As you may have gathered, I'm mightily impressed with Dragon Naturally Speaking 9. So far, in some heavy testing it's done everything I've asked of it, pretty much flawlessly. If it made no mistakes, I'd be suspicious the install CD hadn't shoehorned some little man into my laptop to type the keys from underneath. OK, on occasion he's slipped, but I'd estimate accuracy so far is around 95%. That's not bad!

I've used it with both a Plantronics headset and a top ------of-the-range Sennheiser multipurpose headset equally as well.
 

But there is a fly in the ointment yet to be solved. Initially, this review was to be of Dragon Naturally Speaking version 10, which is said to be even more improved. And so it may be. But despite the best efforts to date by myself, the distributors and technical support, on my laptop with Vista Pro it absolutely refuses to install, getting itself stuck in a Moebius loop during the install procedure.

This isn't to say there's a flaw in the program - it's more likely that I have something odd installed that's conflicting. It's worth mentioning however, so that if you install DNS10 and get the same error, you'll at least know that DNS9 will probably still work.

Nuance, the publisher assures me it's looking into the problem.
  PC Update Summary:

Market:   Anyone who creates lots of text-based
documents.
Rating:   9/10
Price:   From $149
Pro:   Accuracy, easy to use, flexibility.
Con:  Other than v10 install problem, none really.
Contact:   www.nuance.com

Reprinted from the November 2008 issue of PC Update, the magazine of Melbourne PC User Group, Australia

[ About Melbourne PC User Group ]