Re: Free TTS engines

  • From: "Octavian Rasnita" <orasnita@xxxxxxxxx>
  • To: <programmingblind@xxxxxxxxxxxxx>
  • Date: Sat, 22 Dec 2007 12:43:27 +0200

I am using Jaws with Eloquence, and In the menu Options/Basics/Voices/Global adjustment, I have set the rate to 64%.

This is because english is not my native language, but maybe other users use a higher rate.


----- Original Message ----- From: "Andreas Stefik" <stefika@xxxxxxxxx>
To: <programmingblind@xxxxxxxxxxxxx>
Sent: Saturday, December 22, 2007 1:03 AM
Subject: Re: Free TTS engines

People talk quite a bit about how fast they have the TTS engines going.

Would anyone mind posting a couple soundfiles of your screenreader
reading computer code at your "usual" pace? I'm curious how fast folks
typically have it going. I suspect that, in our pilot studies, we have
the TTS engine going way slower than you folks do.


On Dec 21, 2007 2:38 PM, tribble <lauraeaves@xxxxxxxxx> wrote:
Should have read ahead -- thanks Lloyd and all.  I agree.

----- Original Message -----
From: "Lloyd Rasmussen" <lras@xxxxxxxxxxx>
To: <programmingblind@xxxxxxxxxxxxx>
Sent: Thursday, December 20, 2007 10:41 PM
Subject: RE: Free TTS engines

Natural Voices, VoiceText, Loquento and other modern synthesizers are
concatenative, requiring speech segments to be looked up, best match found,
then interpolated in order to be modeled on a particular human voice and
sound natural.  To adjust their speed, you have to do some more
interpolation and matching, as you do for time-scale modification of natural
speech, and this begins to get rough and less intelligible as you go to
twice normal speed or faster.

The older synthesizers, such as DecTalk, Eloquence and the SSI263 speech
chip used by Artic, Braille 'n Speak, Votrax, Accent, etc. are formant
synthesizers. They work on a vocal tract model, and the parameters can be
varied systematically to alter the speed without creating quite as much
distortion or losing consonants. Of course they don't sound as natural, but this is mostly a hindrance to people who don't spend days and days listening
to them, for whom speed is more valuable than naturalness.  The other
advantage of the vocal-tract synthesizers is that they don't require as much
data to be handled, so they work better for echoing individual keystrokes
than the ponderous concatenative synthesizers.  Even on modern computers,
these speed and responsiveness issues are important.

Lloyd Rasmussen, Kensington, Maryland

> -----Original Message-----
> From: programmingblind-bounce@xxxxxxxxxxxxx [mailto:programmingblind-
> bounce@xxxxxxxxxxxxx] On Behalf Of Andreas Stefik
> Sent: Thursday, December 20, 2007 7:34 PM
> To: programmingblind@xxxxxxxxxxxxx
> Subject: Re: Free TTS engines
> Peter says:
> > I have several of these voices.
> > Wonderful sound for reading but to slow for use in a Gui.
> I say
> You talking about the AT &T ones? Can't you just speed them up? I know
> that, on windows at least, there is a slider bar you can adjust. Maybe
> you mean something else, though. Might not be true for every OS, or
> every voice though. I readily admit I'm not an expert on such things.
> Inthane says:
> I have used the AT&T voices and find them to be the best of the TTS > voices > that I have herd, (well, except for there attempt to make one sound > like a
> Scotsman, ouch! LOL),
> I say
> lol, yaa the Scotsman is hilarious. Well, I guess it's a tough
> decision on what to get.
> Thanks for the thoughts, all.
> Andreas

View the list's information and change your settings at

View the list's information and change your settings at

View the list's information and change your settings at

View the list's information and change your settings at

Other related posts: