| ActiveMac | Anonymous | Create a User | Reviews | News | Forums | Advertise |
|
|
|
User Controls |
|
New User |
|
Login |
|
Edit/View My Profile |
|
|
|
|
|
|
|
ActiveMac |
|
Articles |
|
Forums |
|
Links |
|
News |
|
News Search |
|
Reviews |
|
|
|
|
|
|
|
News Centers |
|
Windows/Microsoft |
|
DVD |
|
ActiveHardware |
|
Xbox |
|
MaINTosh |
|
News Search |
|
|
|
|
|
|
|
ANet Chats |
|
The Lobby |
|
Special Events Room |
|
Developer's Lounge |
|
XBox Chat |
|
|
|
|
|
|
|
FAQ's |
|
Windows 98/98 SE |
|
Windows 2000 |
|
Windows Me |
|
Windows "Whistler" XP |
|
Windows CE |
|
Internet Explorer 6 |
|
Internet Explorer 5 |
|
Xbox |
|
DirectX |
|
DVD's |
|
|
|
|
|
|
|
TopTechTips |
|
Registry Tips |
|
Windows 95/98 |
|
Windows 2000 |
|
Internet Explorer 4 |
|
Internet Explorer 5 |
|
Windows NT Tips |
|
Program Tips |
|
Easter Eggs |
|
Hardware |
|
DVD |
|
|
|
|
|
|
|
Latest Reviews |
|
Applications |
|
Microsoft Windows XP Professional |
|
Norton SystemWorks 2002 |
|
|
|
Hardware |
|
Intel Personal Audio Player 3000 |
|
Microsoft Wireless IntelliMouse Explorer |
|
|
|
|
|
|
|
Site News/Info |
|
About This Site |
|
Affiliates |
|
ANet Forums |
|
Contact Us |
|
Default Home Page |
|
Link To Us |
|
Links |
|
Member Pages |
|
Site Search |
|
Awards |
|
|
|
|
|
|
|
Credits |
 |
Movie Maker and Speech Recognition in Windows Vista | |
|
||
| Time: 01:23 EST/06:23 GMT | News Source: *Linked Within Post* | Posted By: Kenneth van Surksum | ||
|
Michael Patten from the PIX team just published a really cool video up on soapbox and the PIX team blog. | ||
|
Write Comment
Return to News |
Displaying 1 through 25 of 445 Last | Next |
The time now is
4:56:19 PM
ET.
Any comment problems? E-mail us |
| #1 By 61 (71.251.77.56) at 10/26/2006 10:49:18 AM |
|
I know from.... experience... that this demo is HIGHLY edited.
The voice recognition in Vista is total crap. |
| #2 By 2459 (69.22.124.202) at 10/26/2006 11:15:55 AM |
|
Maybe it's your mic/audio settings?
Here's a live demo from SpeechTEK: http://video.google.com/videoplay?docid=5549626370637807497&q=speechtek&hl=en There's also third-party footage at YouTube and istartedsomething. |
| #3 By 13030 (198.22.121.110) at 10/26/2006 11:42:24 AM |
|
"Dear Aunt, let’s set so double the killer delete select all."
Enough said... or, perhaps, too much! |
| #4 By 61 (71.251.77.56) at 10/26/2006 11:53:05 AM |
|
Definately not the mic.
Command and control is much easier than dictation with Vistas stuff (with anything, really) as there is a very limited set of commands that you can do, but even then I end up having to say things 3-5 times a lot of the time (this guy had to do it 2 or 3 times at some points). Dictation is just AWWWWWFUL. |
| #5 By 7754 (216.160.8.41) at 10/26/2006 12:13:18 PM |
|
I'm not a fan of voice recognition in general, but my experience with it in Vista has been very good so far. It does take some adjustment, though--you have to get used to the idea that it recognizes phrases much better than a word at a time, so you have to speak as if you're talking to a person (not necessarily a bad thing!). The quality of the recognition itself is amazing, but if you use a lot of acronyms and lingo, you'll have to do some spelling. The control mechanisms are awesome for that, but it's just not as fast as typing that kind of thing.
Overall, I'd say it's like this--if you're a slow typist, the voice recognition is definitely worth using. Even if you're a fast typist (I am), you can probably outpace your typing ability if you have a relatively fast computer and you've adjusted to the concept (and you're speaking "plain English"--don't have a lot of acronyms, etc. as I mentioned--and, most importantly, you know what you want to say in advance... if you work out your thoughts as you go, the controls are still good, but it's quite an adjustment over a keyboard and mouse; I think this may be good in a way--it may help a person think more quickly on their feet?). It could also be useful with a slate-style Tablet PC. On the whole, though, I'll stick to typing. The biggest problem I have is just a practical one--I don't disturb anyone when I type, but people in the office next to me or across the hall may get sick of hearing everything I dictate. |
| #6 By 7754 (216.160.8.41) at 10/26/2006 12:17:31 PM |
|
ch--that was not a voice recognition bug, but a microphone issue on that hardware (and beta software) that has since been fixed. Check out Larry Osterman's blog about it:
http://blogs.msdn.com/larryosterman/archive/2006/07/31/684327.aspx |
| #7 By 415 (71.127.74.113) at 10/26/2006 12:50:49 PM |
| LOL ... what a goob... |
| #8 By 23275 (209.149.207.40) at 10/26/2006 1:43:53 PM |
|
Forget for a moment where the technology is, or isn't... what I like most about this aspect of Microsoft's development is their dogged determination to eventually make it work.
No matter the subject or venue, Bill Gates and other Microsoft execs at least mention "speech" They keep chipping away at it and they do make progress. As the science moves into multi-cores and the newer tools to leverage what a PC will be able to do over the nest 10 years, it seems very likely that they will succeed. Getting back to the present state, it takes work and of course, a great mic. We've tested and now use an array mic that is powered. Many home help robots use the same model. When used with Vista RC2, its performance is amazing - it can follow one as one moves about a room. Again, it does take work and it does take time to perfect and adjust how a machine will interpret what the speaker says; however, with a little work, the results can be pretty astonishing. We'll be posting some videos of this in use with a six panel control system we built as soon as we finish up our new sites and their community center. I think many people will be surprised at how accurate speech has become in Vista. This post was edited by lketchum on Thursday, October 26, 2006 at 13:44. |
| #9 By 13030 (198.22.121.110) at 10/26/2006 2:28:32 PM |
| #6: Yeah, I read about it being a bug or a microphone problem or a new moon or the presenter had all three biorhythms bottomed out. The simple fact of the matter is that it failed spectacularly. When you make a presentation, you have a duty to know ahead of time that it go well... no ifs, ands or buts. If, for some unforeseen reason, there is a problem, you have a responsibility to have a viable backup plan. If it was a microphone issue, as you seem to believe, then how the heck did it very clearly hear the phrases "delete" and "select all"? |
| #10 By 13030 (198.22.121.110) at 10/26/2006 2:35:29 PM |
|
#7: LOL ... what a goob...
And... he... talks... like... a... ro-bot... ev-en... when... he... is... not... us-ing... speech... rec-og-ni-tion. This post was edited by ch on Thursday, October 26, 2006 at 14:36. |
| #11 By 2459 (69.22.124.202) at 10/26/2006 3:27:06 PM |
|
#9, it wasn't a mic issue, but a number of software issues acting in concert (one of them being intermittent quirks with controlling the mic's gain). Even with those issues, the demo didn't "fail spectacularly". Much of it still worked, but many just played up the bad parts.
http://www.youtube.com/watch?v=kX8oYoYy2Gc |
| #12 By 13030 (198.22.121.110) at 10/26/2006 3:51:01 PM |
|
#11: ...the demo didn't "fail spectacularly". Much of it still worked...
How much tolerance would you have for a keyboard that registered the wrong key press even 1% of the time? How about every 1 in 10000 key presses? I can't estimate the number of words I speak and listen to on a daily basis that are clearly and unequivocally understood. By my standards, a low double digit recognition percentage, even part of the time, is a failure. The system coming up with entertaining interpretations for what was actually said made it a spectacle. Speech recognition is an inherently difficult application. It's not just a matter of hearing the words, but hearing the inflection and understanding the context. These capabilities are more in the realm of AI. A child can hear and understand the difference between "help!" and "help?", but this is not a simple process. It is something that takes many months of exposure, beginning with hearing the mother while in the uterus. There is a reason why speech recognition has been so slow to develop. |
| #13 By 2459 (69.22.124.202) at 10/26/2006 4:46:18 PM |
|
You're applying assumptions about recognition failure to a demo handicapped by bugs in the code. You can draw little about the recognition results of correctly functioning code from the demo.
In the general case, if you're expecting 100% perfect recognition, you're expecting too much. It will be years before the technology gets to that point (if ever, as humans don't even achieve perfect recognition), and currently no comparable system achieves this goal. This post was edited by n4cer on Thursday, October 26, 2006 at 16:52. |
| #14 By 8556 (12.210.39.82) at 10/26/2006 8:38:15 PM |
| Computer, aim phasers at Redmond. Fire! |
| #15 By 47914 (24.225.231.107) at 10/27/2006 10:45:58 AM |
|
#13 if ever, as humans don't even achieve perfect recognition
Exactly, just do the "pass the secret" game with a dozen people whispering to the next person in the line "the secret" and watch what said to the first person in line and what was said to the last person. Criminal lawyers deal with this when multiple witnesses here the same thing, but each has a different version of what was said. |
| #16 By 13030 (198.22.121.110) at 10/27/2006 2:16:41 PM |
|
#13: You're applying assumptions about recognition failure to a demo handicapped by bugs in the code. You can draw little about the recognition results of correctly functioning code from the demo.
I'm just calling like it is. #15: Exactly, just do the "pass the secret" game... Better yet, how many words or phrases do you have to repeat during your daily speech in order for your listener to comprehend what you're saying? None, one, perhaps two. I thought so. Speech recognition, especially of the "double the killer delete" type, is not yet up to the task. I still find it intriguing that the "demo problems" didn't affect the ability for the recognizer to hear "dear", "delete", and "select all". Although, it blew the second aspect of speech recognition I mentioned earlier (context) with those last two phrases. |
|
Write Comment
Return to News |
Displaying 1 through 25 of 445 Last | Next |
The time now is
4:56:20 PM
ET.
Any comment problems? E-mail us |
