RE: Speech Recognition

[Gleef <gleef capital net>]

On Wed, 9 Dec 1998, Paul Dorman wrote:
> Isn't there a rumour floating about regarding Lotus Notes and Domino being
> released as open source next year? And doesn't one of these packages include
> speech recognition?

Ooooh, serious game of telephone here.  I think I read the article that
started this rumour.  The article said that Lotus was contemplating
porting Notes, Domino and SmartSuite to Linux.  It further quoted someone
from Lotus saying that they were thinking of making SmartSuite open
source.  I strongly suspect that the quote was either out of context or
the person (I think a VP) was misinformed, but I would love to see
SmartSuite Open Source.  Also, Lotus makes so little of its income from
SmartSuite that it would not cause major upheaval to switch, so it may
actually happen.  Notes and Domino are another issue entirely, I don't
think those will go Open Source anytime soon.

Regardless, SmartSuite does currently include IBM's ViaVoice.  If it goes
Open Source, IBM could possibly open ViaVoice too.  Don't hold your breath
though.


> I definitely believe speech recognition, speech synthesis, and things like
> ThoughtTreasure should become components of the OSS world. 

There already is an excellent Free speech synthesis program, called
Emacsspeak, find out more about it here:
  http://simon.cs.cornell.edu/Info/People/raman/emacspeak/emacspeak.html

I hadn't heard of ThoughtTreasure until you mentioned it, but there is
an OSS system very much like ThoughtTreasure that is already GPLed.
It's called ThoughtTreasure :-).  Check out:
  http://www.signiform.com/tt/htm/ttlegal.htm


> Imagine being able to simply ask your computer when your meeting was
> with such-and-such, and have it give you an appropriate verbal
> response... The funny thing is, the biggest challenge at the current
> time is getting these three _current_ technologies to work together
> in some kind of elegant way - they already exist in a usable form!

I would suggest hooking them all up to Emacs for starters.  Emacs is
highly extensible that way, and would handle much of the work for
you.  Also, then we could have not only Emacspeak but also Emacslisten
and Emacsthink :-).  The code could also be reused in something GNOMEy (GNOMEthink, anyone?).

> Can you imagine the difficulty with which a business like, oh, I
> don't know, one of those big software companies that make
> [dis]operating systems, would have in creating such a system? I'm
> sure it would be a considerable coupe for the OSS community to build
> that one.

I'm sure that IBM has already done much of the work in such things,
their AI and Speech labs are very impressive.  Wouldn't it be nice if
we could beat them to market, though.  Most of the pieces are there, we
would just need to improve the speech recognition, and tie all the
pieces together.

-Gleef