It would be interesting to see how a flawless voice recognition system would impact bloggers.
Imagine having every word you say logged in text... Google would love you!
Facey Spacey Web Development
· 1 year ago
I wouldn't discount voice recognition so quick. The technology involved is the same technology that makes Google's secret sauce: distributed calculations. This is the ability for them to plug in more and more servers and for their search queries or adsense to be pulled based on instantly doing millions of calculations perfectly distributed over whatever server nodes they have available, to put it plainly.
This sort of tech is what companies are looking to apply to voice recognition. So, essentially to more accurately convert your voice to text, the software must cross-reference your voice to a million other voices and text results. And somewhere in there figure out statistically what the actual text result the majority of speakers were trying to achieve.
Paul, have you used simulsribe? It's surprizingly accurate. Currently their monetization model is a paid subscription, but there's no reason that ads can't be inserted and generated the same way, or even generated more simply based off the generated text.
I think the point is that tons of calculations and database entries have to be analyzed in an instance. Companies like Google and symantic web companes like Twine might have all these cool pseudo AI functions, but ultimately, like I said above, their secret weapon is the power to do all the calculations required of these functions. Which requires computing power and intelligent calculation distribution functions. The actual AI logic is probably a lot simpler than that stuff.
I do forecast successful growth in this sector. However, moreso, I forecast major growth in the distributed calculation stuff. Everyone's talking about the Symantec web. People should really be talking about the underlying technology. I personally want to build this underlying technology and sell it to the myriad startups that will want to do the next big Symantic Web thing, just like all the guys who've been doing web 2.0 startup after another.
Unfortunately, Google will probably beat me, and probably already has with offering such tools to big science facilties.
Anyway, I'd like to hear more about this stuff on mashable. Good stuff tho.
-James from FaceySpacey.com, Your One Stop Social Media Shop
Paul Houle
· 1 year ago
Looking at the CS literature, I can say that the technology is going in the right direction. Researchers have built excellent speech recognition systems that run on small parallel clusters of commodity computers -- these could be economically ported to mobile devices with a little bit of codesign between hardware and software... I imagine a "perception engine" that's a bit like a GPU or the PS3 CPU.
There's also the issue that small devices are capable of doing more and more -- the tiny space for input becomes a big problem.
Voice input has been "the next big thing" for years -- there will be a year that it really hits, but will it be now?
All Comments
Imagine having every word you say logged in text... Google would love you!
This sort of tech is what companies are looking to apply to voice recognition. So, essentially to more accurately convert your voice to text, the software must cross-reference your voice to a million other voices and text results. And somewhere in there figure out statistically what the actual text result the majority of speakers were trying to achieve.
Paul, have you used simulsribe? It's surprizingly accurate. Currently their monetization model is a paid subscription, but there's no reason that ads can't be inserted and generated the same way, or even generated more simply based off the generated text.
I think the point is that tons of calculations and database entries have to be analyzed in an instance. Companies like Google and symantic web companes like Twine might have all these cool pseudo AI functions, but ultimately, like I said above, their secret weapon is the power to do all the calculations required of these functions. Which requires computing power and intelligent calculation distribution functions. The actual AI logic is probably a lot simpler than that stuff.
I do forecast successful growth in this sector. However, moreso, I forecast major growth in the distributed calculation stuff. Everyone's talking about the Symantec web. People should really be talking about the underlying technology. I personally want to build this underlying technology and sell it to the myriad startups that will want to do the next big Symantic Web thing, just like all the guys who've been doing web 2.0 startup after another.
Unfortunately, Google will probably beat me, and probably already has with offering such tools to big science facilties.
Anyway, I'd like to hear more about this stuff on mashable. Good stuff tho.
-James
from
FaceySpacey.com, Your One Stop Social Media Shop
There's also the issue that small devices are capable of doing more and more -- the tiny space for input becomes a big problem.
Voice input has been "the next big thing" for years -- there will be a year that it really hits, but will it be now?