begin  quoting Rick Funderburg as of Wed, Sep 27, 2006 at 02:25:07PM -0700:
> [EMAIL PROTECTED] wrote:
[snip]
> Speech recognition should get somewhat better, but there are limitations 
> when grabbing samples of air vibrations.  Humans compensate for this by 
> analyzing the context, which is something that is very hard for 
> computers to do.

Actually, it's pretty difficult for people as well.  Computers have the
problem in that they don't handle context *changes* very well.

People "correct" what they hear, transparently, and often with amusing
(or distressing) results.  We also pretty poor at saying what we
actually mean, especially when we're distracted.

>                   I assume that one could get much better accuracy by 
> implanting small sensors into the muscles that are involved in speech.  

Heh. Bugger that. Keep your electrodes, thankyouverymuch. :)

> I have had a similar thought for a version of sign language using some 
> sort of glove.  These ideas may help until a direct-to-brain interface 
> is working.

That'll make "WinCE" all the more appropriate.

-- 
_ |\_
 \|


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to