From: cpd@cat.pan.net
"Tim McGaughy" wrote in message
news:Ee6dnd3MXY-4RnrWnZ2dnUVZ_tkAAAAA@posted.toastnet...
> catpandaddy wrote:
>>
>> "RT" wrote in message
>> news:4BE7831A.D8F3332B@hotmMOVEail.com...
>>> catpandaddy wrote:
>>>>
>>>> "Tim McGaughy" wrote in message
>>>> > RT wrote:
>>>> >
>>>> >>> Voice recognition is also actually very reliable. They're not
>>>> >>> things
>>>> >>> that devices can 'barely do'.
>>>> >>
>>>> >> Let us know when they pass a Turing test.
>>>> >
>>>> > That's an unfair evaluation when trying to determine how much
>>>> progress > has
>>>> > been made. You expect the performance of a finished product from a
>>>> work > in
>>>> > progress.
>>>>
>>>> I would and still do maintain that voice recognition systems have
>>>> indeed
>>>> aced the Turing Test, in that they succeed and fail in similar places
>>>> to
>>>> where human beings also succeed and fail. Ever meet a human being who
>>>> never
>>>> mis-hears something?
>>>
>>> Do you have a cite or three that such acing has occurred?
>>
>> It is a trivial observation that human beings and speech-recognition
>> software both misunderstand spoken words alike, under the same sets of
>> circumstances and for the same reasons.
>
> Do you know what the Turing Test is?
Simply put, without doing a lookup and cut-n-paste, it's whether a machine
acts "close enough" to a human being (in the respect under consideration)
that an observer has a difficult time distinguishing between the two.
I'm using the term in a fast and loose way here, for the benefit of our
fast-and-loose-with-reality friend here. In the case of speech recognition,
I just happen to think it shouldn't be *more* adept than a typical human
being when it comes to hearing poorly pronounced words. ;o)
Just take it as a figure of speech in this case, no pun intended.
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
