28 Oct
1998
28 Oct
'98
12:25 p.m.
Hi Doug and all,
At 04:59 PM 10/28/98 -0600, you wrote:
On Wed, 28 Oct 1998, Sam Ismail wrote:
In a recent "Electronic Design" magazine, the column "40 years
ago", there
was an article about IBM and digitized speech. It is an interesting column,
as well as Bob Pease's.
-Dave
Unless you used "1000 word vocab" to
qualify the claim, in which case I
still think there may be an earlier contender. Speech recongition goes
back a long way, perhaps as far back as the 50s.
Yes, commercial and 1000 word vocab were the qualifiers; let me know if
you still challenge the claim, otherwise it gets immortalized in a web
page soon.
And given that AI didn't really get started until around 1960, and that
storage for digitized speech would have been very expensive back then, I'd
be interested in any references you could provide for speech recognition
way back in the 1950's.
28 Oct
28 Oct
6:22 p.m.
On Wed, 28 Oct 1998, dave dameron wrote:
In a recent "Electronic Design" magazine,
the column "40 years ago", there
was an article about IBM and digitized speech. It is an interesting column,
as well as Bob Pease's.
Can you summarize the article? I can imagine a lot of ways to digitize
speech without a lot of RAM, but I'm curious about what they were doing in
the 50's with the data. I seem to remember things like "formant"
decomposition and analysis to be reasearch topics in the 70's. Come to
think of it, I don't even know when things like the FFT were invented, but
I thought even that was not until the 60's or so.
-- Doug
11:26 p.m.
On Thu, 29 Oct 1998, Doug Yowza wrote:
On Wed, 28 Oct 1998, dave dameron wrote:
I remember IBM did a rather large push in the mid-60's of speech. They
even sent out records (either the flippy kind or real) with both sides of
'Bicycle Built For Two' - one which was used in '2001 - A Space Oddessy'
for Hal in shutdown mode. The first side was current technology until
this latest technology from IBM and the second side was the improved
version. Along with it was a whole ad blurb of how wonderful this would
make the world along with very little technical data.
BC
In a recent "Electronic Design"
magazine, the column "40 years ago", there
was an article about IBM and digitized speech. It is an interesting column,
as well as Bob Pease's.
Can you summarize the article? I can imagine a lot of ways to digitize
speech without a lot of RAM, but I'm curious about what they were doing in
the 50's with the data. I seem to remember things like "formant"
decomposition and analysis to be reasearch topics in the 70's. Come to
think of it, I don't even know when things like the FFT were invented, but
I thought even that was not until the 60's or so.
-- Doug
29 Oct
29 Oct
12:22 a.m.
On Thu, 29 Oct 1998, Brett wrote:
> > In a recent "Electronic Design"
magazine, the column "40 years ago", there
> > was an article about IBM and digitized speech. It is an interesting column,
> > as well as Bob Pease's.
[...]
I remember IBM did a rather large push in the
mid-60's of speech. They
even sent out records (either the flippy kind or real) with both sides of
'Bicycle Built For Two' - one which was used in '2001 - A Space Oddessy'
for Hal in shutdown mode. The first side was current technology until
this latest technology from IBM and the second side was the improved
version. Along with it was a whole ad blurb of how wonderful this would
make the world along with very little technical data.
Are we talking about two different things? It sounds like you're talking
about computer generated speech, which is a much different problem than
automatic speech recognition (for which digitizing the speech is the first
step).
Kurzweil also had a product that produced speech from text (which it
recognized from a page scan) called Personal Reader. I think the actual
speech generation was a licensed version of DEC-Talk.
-- Doug
9508
days inactive
9509
days old
3 comments
3 participants
participants (3)
-
danjo＠xnet.com -
ddameron＠earthlink.net -
yowza＠yowza.com