Dragon NaturallySpeaking 8
In a fit of wanting to type faster, I looked into the latest version of Dragon NaturallySpeaking. After 30 mintues of googling, I’ve decided that speech recognition is still not ready for prime time. :-( (dear reader. Do not dispair. Read my update below!)
I found several blogs and sites that talked about how excited they were at the prospect of how the software could help them, but I never found any followups. That says to me that everyone who tried it got dis-interested quickly. When I first tried speech recognition several years ago, I had a similar experience. After a few days with it, I thought that if I only put more dedication into teaching the computer how I spoke, I could get some use out of it. But I just wasn’t interested in putting that much effort into it. Hence, the waning interest. Well, here we are, 10 years of research, my computer is 200 times faster (!!! 3 Ghz vs 14 Mhz!!!) and the reviews still say pretty much the same thing.
The most useful review was from John Udell’s Weblog. He included a video of him dictating a letter. His was virutally the same experience I had 10 years ago. The recognition had about a 2% error rate. That sounds good until you realize that this posting so far is 203 words… That means there would have been 4 errors in the preceeding text… errors that were spelled correctly and were likely gramatically correct, just not what I intended to say.
So then you have to correct the errors… That can be terribly slow, and error-prone in itself in an audio interface. Listening to John Udell patiently talk to his computer in a carefully moderated voice, and having the machine still make dumb mistakes drove me crazy, and I’m a patient guy.
Grr. I don’t know… Maybe I will give it one try. I tried to type as fast as John was dictating and I very quickly realized that, when it worked well, he was going at like 100 to 120 words per minute. I type at something like 25-40 WPM. I would love to be able to integrate this kind of performance in my typing life! Maybe I could dictate and then edit by hand? I don’t know…
update 4-29-05: I’ve been using Dragon NaturallySpeaking 8 now for two days and I’ve got to say that I’m extremely impressed with it. My previous reservations were unfounded. I’m typing this right now with my voice. It’s pretty darn cool, and yes, I’m going a lot faster than I could type after only one day of training. it takes a little bit of getting used to, speaking to the computer, but really not that much. More importantly, I feel that I’m using a different part of my brain in order to write things. It’s a speaking thing, not a writing thing. That was one thing that I was hoping that I would get out of this.
Woot!
I’ll keep my intrepid readers abreast of how I’m doing with Dragon NaturallySpeaking.
Oh, and as for the microphone, I’m surprised to say that I’m very happy with a cheap lapel mic that came with my web cam.
Gosh darn it, it’s even becoming easier to say things like “Send that” instead of clicking on the Send button in Outlook.
You know, I have to admit that it’s kind of nice to hear my voice in an otherwise quiet room. It’s better than listening to the mindless, brain sucking television in the background.
Update 5-3-05 I’m convinced.
I am now able to type and about 80 wpm. That’s twice as fast as I have ever been able to type in my life. With more practice and the new headset that will be arriving in a few days, I am fairly confident that I will be able to tie at 100 wpm very reliably. This gosh darned thing is good! There are still a few small issues but they all seem conquerable. For example, right now the integration with Firefox is less than perfect. But there are tools to get past that. I’m really pretty impressed.
Of course, instead of me doing just necessary things faster, I am now becoming more verbose. I think I like that in my Internet life.
Case in point: this is getting to be a pretty long blog entry, isn’t it? It’s not that I’m spending more time writing this entry, it’s just that I’m “typing” a lot faster. Woot!
(and it is a bit of a novelty teaching the computer to understand the word “woot”)
update 5-5-05: That’s it, I’m hooked. Dragon NaturallySpeaking 8 is amazing. I really can type/speak at about 80 wpm. It’s still taking a bit of getting used to but darn it, this thing works. I went out and bought a good dictation headset (an Andrea ANC-750) from Knowbrainer.com and that has improved the accuracy quite a bit (of course, it’s also an excellent gamer headset ;-). I’m thinking less and less about how I speak to dictate after just one week.
Update 5-20-05: I continue to be happy with Dragon NaturallySpeaking 8 (DNS8). My biggest gripe with it right now is its memory and performance footprint. When it’s running it takes longer than normal to switch between windows, even if DNS8 isn’t engaged. That makes it so that it’s a bother to leave it running on the odd chance I’ll want to issue a “close window” or a “send that” command. Though I admit that I’m a huge short-attention-span-theater window switcher. I have 2 monitors and at the very moment I have 12 windows open. On any day when I’m spending more time writing than not, DNS8 stays running.
Update 9-14-05: FYI I continue to be an avid Dragon NaturallySpeaking user. I got a DMCA takedown notice from copyright-compliance.com last week representing Scansoft saying about this very page (where I gush happily about DNS)…
It has come to the attention of Scansoft that you are distributing unlicensed and unauthorized Scansoft Products.
If anyone can find the unlicensed Scansoft product on this page, I’ll give them a prize.
Update 10-7-05: Sweetness. I just installed another gigabyte (bringing me to 1.5 GB) and all of the lag that I’ve been seeing when moving between programs has gone away. I can now leave Dragon NaturallySpeaking running much more of the time without a strain on my PC.
The only negative thing I have read elsewhere that seems credible
is that most people are not used to dictating when doing creative
writing.
I’ve found that I do pretty well speaking my ideas into Dragon NaturallySpeaking. I think I do a little bit better speaking them than I do typing them. It has taken a little bit of training to make myself speak clearly… thinking out what I’m going to say before I say it… but I’m getting pretty good at it. I am dictating this entire note and it’s going pretty well, don’t you think?
It is a little weird hearing myself speak in a slightly emotionless way, but I think if I put my full emotion into what I was saying, there would be no hope of Dragon NaturallySpeaking picking up what I’m saying.
You mentioned tools to get Dragon to work with Firefox – could you elaborate?
Oh dear, sorry but I forgot what the tool was… The tool I played with (I don’t recall the name) allowed the user to say things something like “Highlight links”
[all the links would get numbers next to them] “Click 5”. If you find it, tell me, would you?
The tool for Firefox links might possibly be either,
Hit-a-Hint http://users.tkk.fi/~psillanp/hah_hp/
Mouseless Browsing https://addons.mozilla.org/extensions/moreinfo.php?id=879&application=firefox
Yes, Hit-a-Hint. That’s what I had previously found. And now there’s another, wee! I haven’t given either a run for it yet but will try soon.
I have set up a Firefox extension called “mouseless browsing”
It assigns numbers to every link on the page, and it might be possible to use it in conjunction with Dragon NaturallySpeaking, but I have not figured out how to do so.
If it was possible to dictate key strokes you could just say the number and enter, and then you’d be able to navigate without touching a mouse or keyboard.
Please let me know if anyone has tips on how to do this with Dragon.
Hey! DNS does pretty darn good text-to-speech! My eyes started hurting today after reading so much so I tried pushing some wikipedia entries through DNS. While it’s not perfect, I can definitely listen to it. And while “Jennifer” reads aloud slower than I can read, when it goes in through my ears rather than my eyes, I can do other things with my eyes and hands. IE, I can half-listen but I can’t half-read.
Cool.
It’s a slight bother in that I have to copy whatever I want Jennifer to read into a Microsoft Outlook email, select the text and then click Sound | Read That.
Hello… I started using the dsn8 and I would like to know if you are still using it and if you have been making progress… Thanks!
I continue to use it on occation but not as often as I might. It shines when doing long-form entries but not for short stuff. And most of what I do on the computer is short stuff. IE: look at my most recent blog entries. They are all either short or technical enough that DNS wouldn’t have worked well with them.
What keeps me away is the slight inconvenience of putting on a headset, the startup time, and that DNS owns my keypad’s + and – keys.
Page emailed me this comment:
————————-
I use NaturallySpeaking all the time, because I have to. I understand the inconvenience for people that don’t absolutely need to use it.
A couple of thoughts for those who want to pursue NaturallySpeaking-if you have lots of technical/jargon terms, make use of the Vocabulary Editor. Say you have something like “Cr(VI)” (from one of your recent posts) and you want to have dragon write it correctly. Add it to Vocabulary Editor (Dragon Bar|Words|View/Edit) with a different spoken form. In the Written Form Bar|Words|field of Vocabulary Editor, enter exactly what you want Dragon to write: Cr(VI). In the Spoken Form field, enter: C. R. Six. When Dragon hears you say “CR6”, it will write Cr(VI).
Additionally, you can change Dragon’s default keys. Go to: Dragon Bar|Tools|Options|HotKeys and reassign the microphone on/off and the microphone sleep key. I use Pause for the microphone on/off.
Page
To use Naturally Speaking with mouseless browsing, simply say “press” or “click” and then the number.
This is the worst product for any one with a English accent. I tried 7 and 9 and they don’t work. I returned 9 after a few days and stated it never worked and Nuance never bothered to follow up or investigate what the issues were. I don’t recommend at all