Here's an excerpt of a posting that triggered my IMSAI smart agent.
Does anyone recognize his name, or the company he's describing?
- John
Jefferson Computer Museum <
http://www.threedee.com/jcm>
Subject: Re: WRQ's _at_Guard
From: Wirt Atmar <WirtAtmar_at_AOL.COM>
Date: Sat, 20 Jun 1998 15:07:56 EDT
Message-ID: <3dda660e.358c088e_at_aol.com>
Mailing List: HP3000-L_at_utcvm.utc.edu
....
Just because something is possible doesn't mean that anybody's gonna want
it.
The very first business that AICS was involved in was computer-based speech
synthesis. Indeed, that's where our name comes from (AICS = artificially
intelligent cybernetic systems). The company grew out of something that had
previously been just a hobby of mine -- and a potential dissertation project
for graduate school: completely host-based speed synthesis, de novo,
where an
electrical analog of the human vocal tract was commanded to truly synthesize
new speech rather than simply play back digitized human speech.
In 1976, when I finished school and the company was formed, we actually
made a
lot of money surprisingly easily and surprising quickly selling speech
synthesizers as drop-in boards for Altair and IMSAI microcomputers. That
was a
time you could sell anybody anything. Every conference was a feeding
frenzy of
buying. You threw something up in the air and ten people wanted it.
In 1976 also, when I became a professor at our local university, several
students also continued on with the speech synthesis work. Vickie Kurtz, a
person who you actually sort of know because she is more or less single-
handedly putting together QCTerm now, did her master's work with me then.
For
her thesis, Vickie put together the best vowel synthesizer that I ever
heard,
using voltage-controlled oscillators, controlled by a simple rom-based
finite
state automaton so as to simulate the mechanical inertias found in the vocal
tract. Although her synthesizer didn't have any sort of noise generator
in it,
so it couldn't form the fricatives (hiss-like sounds) necessary for full
speech synthesis, its vowels were the most human-like I've ever heard and
were
essentially indistinguishable from a baby's babbling or cat-like animal.
Vickie finished in 1982.
A year or two earlier, Texas Instruments got into speech synthesis in a big
way using a technique halfway between digitized speech and speech
synthesis, a
process that might now be called "smoothed sampling". It worked much better
than true synthesis in that its intelligiblity was quite good. Because of
that, TI was able to sell a variety of manufacturers on the idea of having
talking washing machines, microwave ovens, vaccuum cleaners, etc. The most
expensive of these talking machines was the Chrysler LeBaron automobile
-- and
people (the end-users, not the engineers) found all of this incredibly
irritating. The single most requested option on the LeBaron, by far and
away,
was to have the speech generator deleted.
Talking appliances were a technological fad that lasted only about three
years, 1979-1982.
...
Wirt Atmar
Received on Wed Jun 24 1998 - 21:07:40 BST