Voice Recognition Software (Blog)

[Savage]

Use this thread to discuss the following entry from Steve Pavlina's blog:

Voice Recognition Software

[Jeff Lilly]

I'm glad you're finally having a good experience with voice recognition software, Steve. I've been using Dragon for almost three years now, since I joined the company, and it's been wonderful. For what it's worth, our internal testing shows Dragon to be considerably more accurate than Vista, with no training necessary at all. Of course, Dragon doesn't come packaged with Microsoft's operating system.

[Dan.Linehan]

I'm pretty sure when Google launched their free 411 service (800-GOOG-411) they were mainly seeking to accumulate a massive dataset to use for the development and evolution of their own voice recognition software. Not to mention the giant sandbox for testing.

It'll be interesting to see what they do with it.

[CannibalSmith]

How do Vista and Dragon handle foreign accents? When I tried using Windows XP voice recognition I had to make a tremendous effort to pronounce "th" (as in "the") correctly (instead of "da" or "ze") for the software to achieve any accuracy.

[Jeff Lilly]

An acoustic model could be built to handle your accent, CannibalSmith. However, the model wouldn't work very well for speakers of standard English. :-)

Voice recognition software always does a lot better if all words are pronounced clearly and crisply, at a normal speaking speed. There is still a lot of work to be done before the software will be able to handle both accented and non-accented speech easily.

[RealEstateMegaBook]

Dragon can handle almost anything you throw at it. The latest versions are exceptional, requiring little or no training to achieve accuracy. It even works with multiple languages and different accents on different languages. For instance, they have different vocabulary sets for US and UK English.

Like Steve mentioned in the article, I'm using it because I'm disabled and unable to use my hands at all. I'm using it to write this post right now, and I can do pretty much anything I need to on the computer with it. In fact, I can even change the channels on my TV, answer the phone, make calls, or whatever else I need, using nothing but my voice.

It's a lifesaver... literally :-)

[BChris]

I tried it and its very decent but obviously not perfect but the point is that this kind of software is improving really fast and it will be interesting to see it replacing the keyboard at all. What is your opinion about this software replacing the keyboard in the future Steve?

And by the way since you guys are talking about dragon where can I download it to try. Is it free?

[Calculusaurus]

Just curious: what's your typing speed?

I type pretty fast, but I can't keep up with the speed of my thoughts, so typing papers can often be a tedious process of micromanaging each phrase. It's like I have to stop every few words and say "wait, where was I going with this again?" But in conversation I can string thoughts together fluidly (like most everyone else), so maybe using speech recognition software would actually be a huge improvement in stage 1 of writing -- spilling your thoughts down on paper.

[CannibalSmith]

Originally Posted by BChris

Is it free?

No, it's 100 bucks, and they don't seem to have a trial version available for download.

[Lenie]

Thank you for the reminder, Steve.

I'd forgotten that it came with Vista...

[Jake Birkett]

yeah I've been trying voice recog software out every so often for about 10 years or so and only recently have I found it good enough to actually be practical (I use IBMs Dragon thing), although it is still a bit slow - so I'm hoping it'll be better when I get a new PC.

I write lots of emails to my team members when I writing a game and it's been useful in reducing the time spent typing, however I still need to reduce the actual amount of emails required ;-)

[ogrekilleat]

I would like to know how many wpm you can speak.

[MasterD]

You might like http://www.copytalk.com

[openeyes]

Originally Posted by ogrekilleat

I would like to know how many wpm you can speak.

With Dragon Naturally Speaking version 9 it's said to work at up to 160 wpm. While it may be a little trickier, with the right digital audio recorder (which is specified on their site) you can also take little audio notes while you're away from the computer, plug it in when you get home, and have it convert the recording of your voice to text.

The main drawback I see to it is that many people are in the company of other people when they're typing something, and they may not feel ok voicing what they want to write in that situation, or at least the people around them may not want them to. Having an office with everyone using one (or a library or dorm room) could make it sound somewhat like a call center. It could be fine when the person is alone though.

[Savage]

I type at 60wpm but can easily speak at triple that rate.

[Lenie]

60 wpm + triple?

[Andreas]

Just to clarify for Lenie, Steve means that he can speak 60x3 wpm, which is 180 WPM, which means that Steve could be certified as a Fast Talker™.


Steve, you would tell us if this was a post sponsored by Microsoft, right? I think so.

[Savage]

Originally Posted by Andreas

Steve, you would tell us if this was a post sponsored by Microsoft, right? I think so.

I don't do any pay-per-post stuff, so there's no sponsor. I just happen to like the software.

[Savage]

I am a fast talker. I can usually slow down to 150wpm, but below that rate I risk putting myself to sleep.

[Trustme]

I had to learn how to type back in school years ago. I hated it but I am glad they forced me to do it. I can touchtype at an amazing speed today and if you really want to enhance your productivity this way it's much smarter IMHO to actually learn proper typing instead of experimenting with this kind of software. It's a skill that can be learned pretty fast - a couple of weeks with daily training and you should reach a point where you never even consider speed recognition software.

[Dylan]

Dear Steve,
I just navigated to the site and typed this without touching the keyboard. A few hiccups, but overall, pretty cool. The quote show numbers quote feature is especially interesting.

[Keith]

Sounds good.

From some of the other poster's comments it sounds like Steve's improved results are due to recent improvements in the technology rather than differences in brand. (Steve, I'd be interested to know when you last used Dragon?)

Being bundled with Windows is hard, but not impossible, to beat. I'm entering this post with Firefox after all.

[Neutral]

What about the loony bin effect?

I'm still not used to people who talk with blue tooth pieces...those spacemen!

[elliot]

One noticeable piece of information Steve didn't really expand upon are the options available to Mac OS X users. Since the 90s, iListen has been the primary program Mac users relied upon for speech dictation needs. A partnership existed between iListen parent company MacSpeech Inc and Phillips that created software that worked well but involved substantial amounts of training.

MacSpeech is now aligning itself with another company, Dragon Systems to essentially bring a version of Naturally Speaking for the Mac. This has been desired for years by many, including David Pogue of MacWorld / New York Times / Missing Manual fame.

The software is being released later this month and I have already pre-ordered it.
Here is a demonstration on youtube (YouTube - MacSpeech Dictate - apresentação na Macworld Expo 08)

While not built-in, it certain seems like a viable alternative for people who don't want to type all day.

[Crash]

Hi Steve,

since you're having problems with Vista's VR-software not supporting non-MS-software, you may want to check out Microsoft Live Writer, an offline blogging software that also supports WordPress.

[Lenie]

Originally Posted by Crash

Hi Steve,

since you're having problems with Vista's VR-software not supporting non-MS-software, you may want to check out Microsoft Live Writer, an offline blogging software that also supports WordPress.

What's next? Let me guess: you people have all beta-test drove Silverlight too?

[Shagaia]

Is it possible to train either program for two voices so that the two people can carry on a conversation and capture it? Thanks

[Gene]

Originally Posted by Crash

Hi Steve,

since you're having problems with Vista's VR-software not supporting non-MS-software, you may want to check out Microsoft Live Writer, an offline blogging software that also supports WordPress.

I recently moved to Live Writer for this very reason. I use WordPress and have used other off line tools before trying the Vista voice recognition software. It works for me.

Having lost a portion of my left hand in an injury I type slowly at less than 45 words per minute with frequent errors at best and must look at the keyboard rather than the screen since without fingers I have no way of finding where my hand is on the keyboard by touch.

With Live Writer and Vista VR I can devote more attention to the flow of the article. I easily get several times the word entry speed from VR compared to manual typing. I need less editing to correct errors and the flow of writing is easier to enter and maintain.

[CannibalSmith]

Originally Posted by Crash

since you're having problems with Vista's VR-software not supporting non-MS-software, you may want to check out Microsoft Live Writer, an offline blogging software that also supports WordPress.

But it doesn't support LiveJournal. Boo.

[DuffOMelia]

I'm surprised you're using Vista rather than a Mac. Vista to Mac : now that's a productivity increase.