Re: dtk-soft

[Doug Smith <bdsmith@xxxxxxxxxxx>]

That's right, that's exactly how it works. To prove it, do this: 

Use any algorithm you can develop and re-record the sound of the voice
from outloud. Now, if you recorded it in analog, re-digitize it so we
can do the next step. If you recorded it in straight-up digital, skip
the re-digitization process. In the next step, use whatever algorithm
you can develop to separate the sound of the voice into its component
parts. This will reveal the kinds of sounds used to create the various
sounds in the words. If you space the components far enough apart, you
can hear what's going on. I can hear it because I am accustomed to
listening very carefully to sounds. 

As a blind research computer scientist, I am interested in speech
synthesis for 3 reasons. In fact, I will, soon, be working on a
project to tie speech synthesis and voice recognition together to
create an integrated system of the two which will require, hopefully,
no training by the user in order that it might recognize voices. The
reasons for my interest are: 

1. Speech is how I access the computer which is my entire collection
   of scientific instrumentation and communication with the outside
   world. 

2. determining how speech synthesis really works so I can,
   prayerfully, create a product that blind people the world over will
   want to use. 

3. The extremely fascinating prospect of being able to create a speech
   system that can work, using the same knowledbgebase to both speak
   and recognize speech, again, very prayerfully, with no training by
   the user. 

Where do I get my documentation? I haven't had the chance to pull down
much from the net right yet, but a bit of googling should track down
enough to keep you, me, and the rest of all blind computer scientists
busy for the next millennium or so. I also get a great deal of it from
picking apart source packages. I will download a source package and I
have no more the intent to install it than I do to change operating
systems. I will sit here for hours and pick apart the source code,
analyzing it line-by-line until I understand, at least, the general
algorithm the author used. Then, as a good researcher should, I take
this information and look through it to see where I might be able to
improve. After this, I start coding little experiments and gradually
build them up into big experiments and finally into pseudocode and a
completed package. 

I am just about to embark on another collaborative project with a
fellow member of my Linux Users' Group, and I have a couple more large
to medium-sized projects I need to get out of the way then, if you
wish to work with me, we might be able to "talk turkey" on this speech
system. Write me back if you're interested. 

Well, I guess I had better get this done and finish my mail. I want to
read more on what happened to the shuttle. By the way, I guessed right
on what caused the crash. It was that piece of insulation that flew
off the tank and broke up the tiling on the port wing. They should
have landed. The video showed it clearly, the newsman said, and
there's no way in the universe they could have analyzed it from the
ground. They should have gone out over the Gulf of Mexico, jedisoned
the power package, which includes the tank and the solid fuel engines,
circled back around on the main engines, lined it up with the runway,
shut it off and landed. That mission could have waited. 

Does anyone know about the prospects of becoming the first blind
person in space? I'm ready. 

Well, got to go for now. I hope you will write me back. 

-- 
Doug Smith: C.S.F.C.
Computer Scientist For CHRIST!

-----------------------------------------------------------------------------
To unsubscribe from the emacspeak list or change your address on the
emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
subject of "unsubscribe" or "help"