Photokina 2014 highlights

Dragon NaturallySpeaking 10 speech-to-text engine reviews itself

By

July 1, 2009

Dragon NaturallySpeaking 10 Preferred Wireless: an honour to goodness wife shaver if you h...

Dragon NaturallySpeaking 10 Preferred Wireless: an honour to goodness wife shaver if you having rubble typing.

Note: this entire article has been written using the uncorrected output of Dragon's NaturallySpeaking speech-to-text engine. Regular podcast listeners will know that a couple of weeks ago I had the misfortune of falling off my motorcycle. Well, it turns out I broke my wrist in the accident, which makes it pretty tough to type. This stupid plaster casts can be on for at least six weeks so I figured I'd better watch out an alternative that would let me keep writing. And here it is: the number one speech to text engine on the market, Dragon NaturallySpeaking. And what better way to review a speech to text product and the Post an article written entirely using it, and completely uncorrected will stop

Last time we looked at the Dragon speech to text engine it was pretty rudimentary. But that was several years ago, and things are bound to have come a long way since. Now that I'm forced out of necessity to look at hands-free typing solutions like this, the stakes are a bit higher will stop

The first version of Dragon Dictate sold to US$9000 per single user licence back in 1990, which puts version tends 450 Australian dollar prostate into some context. By prostate, of course, I mean pricetag. (That price is for the preferred wireless version, which includes a Plantronics Bluetooth headset stop)

Installation and setup is simple enough, you get to choose which language dictionary you're going to translate into and you get to warn about what kind of accent you've got. In my case that some of the accent -- sorry, in my case that an Aussie accent, but the system can cater for British, Indian, Spanish, Southeast Asian, and several different types of US accents will stop

Choosing the right accent, dictionary and audio device set of options are of critical. The software was completely unable to understand me if I, for example, used to Bluetooth headset with the audio levels were set for a line-in microphone will stop -- by the way, every time you see "will stop" at the end of a sentence, that's just Dragon mishearing me saying "full stop will stop"

This is day two of my Dragon experience, and I understand that the system needs to spend time learning and adapting to the way that I speak. Each time, for example, that I correct a word that's come out wrong, Dragon apparently takes this into consideration the next time you something similar.

There is also the option to "train" the system by reading it a selection of texts to allow it to calibrate your voice type. I've already done three have about 12, but I'm not sure the unseen massive improvements.

I should also add, that as a professional writer I'm very much used to thinking through my fingertips. It's actually proving to be quite a challenge to learn to construct sentences verbally and on-the-fly will stop it feels clumsy and awkward -- and I find that to be a really interesting experience, a new skill to get my head around.

In terms of user interface dragon is fairly flexible. V "DragonBar" can sit quietly in the window on top of everything else, all docks to different parts of the screen depending on your preferences will stop is the number of voice commands to let you use context menus and what not around your operating system, but frankly I'm a bit frightened of that functionality; simple correction commensal years are still kind of hit and miss, so I don't Wanna gives Dragon access to buttons like "file -- save" just yet.

Where Dragon does Miss EU and take down the wrong word, it's easy enough to go back and corrected, but this is a pretty fiddly process. In fact, the claim on the box that writing documents using Dragon is more than three times faster than typing, is looking like absolute bushes to me. Getting a slab of text down on the page might be a lot quicker, but the process of going back on correcting every mistake using voice commands seems to take forever will stop

One thing it has proved exceptional at his amusing friends in instant messaging conversations. Nobody seems to get tired of listening to me complain about the sucking stupid thing, and while I'm aware that you can train to swear, I find it much funnier to see what the car comes up with by itself stop

I've got at least a month or 2 to spend getting used to Dragon, and I have no doubt that it will improve as it says it will on 10. We record a lot of interviews here it Gizmag, and were looking forward to the opportunity to use it to transcribe interviews for its. Note: the ability to import audio files and have them transcribed is only available on the preferred and professional editions, not the standard edition of Dragon 10.

But I have to say that at first glance, it seems that the biggest surprise about speech to text software in this day and age is that there doesn't seem to be a lot of surprises. if you have ever spent time on the phone yelling and swearing an automated response system, you know roughly what you're in for will stop for the most part, Dragon seemed to do its job fairly well, but when it's stuff things up its sucking infuriating.

Perhaps in the interests of fairness, I should post another review in a month's time. Till then apologies in advance to my editor Knoll is perhaps on correcting the -- wow, way to end on a bad night.

About the Author
Loz Blain Loz has been one of Gizmag's most versatile contributors since 2007. Joining the team as a motorcycle specialist, he has since covered everything from medical and military technology to aeronautics, music gear and historical artefacts. Since 2010 he's branched out into photography, video and audio production, and he remains the only Gizmag contributor willing to put his name to a sex toy review. A singer by night, he's often on the road with his a cappella band Suade.   All articles by Loz Blain
5 Comments

To be fair you have to slog through ALL of the training "thingies" (technical term) before it really begins to work well. If you write technical/engineering articles as I have then you also have the joy of adding in a custom dictionary and training it to understand those. After all of that it is remarkable how well it does. Not being able to type that fast I found it was wonderful to be able to just express my ideas or concepts aloud as fast as I could conceive them instead of type a bit and then wonder if that was still the same. It's a wonderful tool but you have to take time to sharpen it.

Wragie
1st July, 2009 @ 01:41 pm PDT

I use Dragon regularly (10.0 medical at work, consumer at home). I find it indispensable and a tremendous upgrade from past versions (and other voice recognition programs). The only disadvantage is that I can't dictate while someone is in the room with me and have to downgrade to typing so that I can carry on a concurrent conversation. (luckily it gives me an excuse to shoo the family away)

Thomas George Lareau
1st July, 2009 @ 04:05 pm PDT

Whilst the capabilities of the software are on display during this review, it's comical ineptness does make for a great and interesting article! I cannot confess to having a need for this sort of product unless i was to take up the vocation of comedy speech writer, but having been made more aware of how these things are supposed to function (or more succinctly, how they don't) i'm warming to them more.

Thanks go to the writer for such an enjoyable read, to the editors for letting it through, and to the software developers for their pure guts and sense of humour !

sharkuss
1st July, 2009 @ 08:25 pm PDT

"absolute bushes" - classic!

David Fromant
2nd July, 2009 @ 12:04 am PDT

I must admit to being disappointed. My adult niece is visually impaired although she has gained her masters and is a national political appointee in our country. Ten years ago I tried MS speech to text engine and found that I had much the same complaints about it that I had with the very first version of Dragon. I'm disappointed that I won't be able to report to my niece who types text into her computer using her braille keyboard that there is now a reliable product out there to relieve her of that tedium and, more importantly, she has access to the national dialog on such topics.

I also have this vision of "conversing" with my computer. Actually I mean requesting information from Google and YouTube, about which I just made the following post:-

"I would have returned to University study years ago but for the existence of YouTube and Google. These have the potential for doing more about the information gap of the world than any other invention since the printing press, including universities. Google and YouTube have answered in detail every question that I've had for the last several months."

The frustrating thing is that it seems that the STUPID developers have simply not realized that the biggest impediment to adoption of their product is not simply its reliability but equally the ease of making corrections, especially for people who cant see the keyboard. This problem seems not to have been relieved in the slightest, since Dragon first started. HOW DUMB IS THAT!

All that is needed is FIRST of all, the product needs to be integrated with a text to speech converter, if it already isn't. SECONDLY: A line counter has to be integrated in the product that can be turned on and off. THIRD: A numbered word counter should also be integrated that can respond to an instruction like: "Correct Line 5 third word on, "or "third word to end of line," or "third word to (punctuation) mark." It shouldn't be necessary then to correct text until the end of, say, a page of direction. FOURTH: The program should call the page endings if required to do so, turn off and turn on. In correction mode then, it should be easy enough to correct a piece of text either to the next punctuation mark, or to the end of the line, since the punctuation mark mirrors natural speech patterns and, conversely, end of line (EOL or NL) are part of the ASCII codes natural to the computer.

Brerlou L. King
8th January, 2010 @ 11:31 am PST
Post a Comment

Login with your gizmag account:

Or Login with Facebook:


Related Articles
Looking for something? Search our 28,547 articles