[CoMoDev] Re: Text to speech engines

  • From: "Andrei Podoplelov" <andrei@xxxxxxxxxx>
  • To: <comodev@xxxxxxxxxxxxx>
  • Date: Thu, 14 Oct 2004 09:50:49 -0600

My last serious approach to Speech technology was probably in 2001. I
tried every available for free trial TTS and Speech recognition engines
for Windows I could find. My impressions were:
1. TTS work pretty good, but ... just one example. Remember Tom Soyer? I
wanted my program to read it for my children. 
The text was:
"TOM. No answer."
We heard:
"Tee-Ouh-eM. No answer" - What would you children think about aunt
Polly? :)
2. Speech Recognition demanded extensive training and even after that
could be useful in command mode only. Also it demands a good processor
power and the mentioned Russian company don't offer it for PPC.
I believe Speech Recognition for Russian is much easier then for
English, because Russian is kind of WYSYWYH - What You See Is What You
Hear language with direct letter-sound correspondence.
Has the TTS/SR landscape really changed during last three years?

-----Original Message-----
From: comodev-bounce@xxxxxxxxxxxxx [mailto:comodev-bounce@xxxxxxxxxxxxx]
On Behalf Of David Beers
Sent: Wednesday, October 13, 2004 6:15 PM
To: comodev@xxxxxxxxxxxxx
Subject: [CoMoDev] Text to speech engines

I've often thought that text-to-speech was a good stepping stone toward
the next mobile killer app: speech recognition.  (What does that make
it, a "stunner app"?  No! sorry I said that to a group of computer
geeks!  Please just recite the Monty Python parrot routine quietly to
yourself until the feeling goes away! ;-)  


Seriously, it would be nice to have your email, an ebook, or a web page
read to you while you are driving, and to have browsers and other apps
written with voice feedback to help you navigate menus without looking
at the screen.  On your smartphone you'll be able to dial from your
address book without looking at the phone or get a verbal caller ID in
your headset instead of having to pull the phone out of a holster or
pocket and look at the screen.  Put this together with even *very*
minimal voice recognition (like the ability to understand a vocabulary
of just a few navigational terms--"yes," "no," "next," "back,"
"address," letters of the alphabet) and you could have hands-free smart
phones (or think industrial devices) that you operate completely through
a small, bluetooth headset.  Powerful stuff, no?


I think we're just about there, now, and if I didn't have so darned much
billable work to do I'd love to be working on this.


Here's a link to a new text-to-speech engine for PocketPC:

Sounds very reasonably priced, which makes me wonder how good it is.  


Here's are some text-to-speech products for Palm OS devices:






The second one is free, for Pete's sake, but neither seems to have a
real API for incorporating the functionality into your programs in a
flexible way.  Gotta be something like that out there, just haven't seen
it yet. Anyone else?




David Beers

Pikesoft Mobile Computing





Other related posts: