Re: forward masking (was: Re: Hugos)

From: Michael M. Butler (butler@comp-lib.org)
Date: Mon Dec 18 2000 - 10:21:12 MST


The problem with "Speedtalk" as a model is that the polyagglutinative
inflected (I think I've got those terms right, but I'm probably kidding
myself--it's morning, before my coffee, and I don't DRINK coffee (:))
language Papa H. outlined, one phoneme per word, one word per sentence,
was still using our legacy audio I/O hardware in our legacy noisy
environment.

My prediction, the first time I re-read that story ("Gulf") after
acquainting myself with Claude Shannon's work on information theory, was
that you'd better be in a damned good environment, or you'd *have* to
switch back to something more redundant like English. Even wind noise in
your ears on a beach might drop your recognition rate off a cliff. And
forget about a cocktail party. Or so I suspect.

But hey, they were all H. superior anyway, so they probably had better
hearing, too. :)

Now, fix the channel noise problem.

KPJ wrote:
>
> It appears as if Michael M. Butler <butler@comp-lib.org> wrote on AI language:
> |
> |As long as they're not using Heinlein's "Speedtalk" from the story
> |_Gulf_. :)
>
> How so? Would not humans also enjoy a CPU and/or communication speed increase?



This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:50:37 MDT