New Text-To-Speech -- Wow!



As you probably know, "The Harrow Technology Report" is always available
as a spoken Web-based "radio program" at a link specified in the "Listen
to This Issue" section.  Several times over the years I've experimented
with the best "Text-to-Speech" conversion programs to see if my voice
had become superfluous, but the robotic-sounding results always sent me
scurrying back to my microphone.  I will probably still continue using
my voice to record these shows, but for the first time I've heard an
alternative that sounds so good that there is now an acceptable
alternative.  It's called Loquendo, a spin-off of work done by the
research folks at Telecom Italia Labs.

Loquendo runs on a PC and, depending on the configuration, can speak a
particular text file.  Or under program control, it can act as a
text-to-speech server providing multiple simultaneous text-to-speech
streams.  Loquendo also comes with quite a few voices, each one
optimized for speaking the text of a particular language.

If you're at all familiar with the many text-to-speech converters that
have been available on the market, you haven't (yet) heard anything new.
But when you actually hear the result, if you're half as impressed as I
am, you'll agree that Loquendo will open up quite a few doors.

This is not to say that Loquendo is perfect.  It's not.  It doesn't
always get things exactly right, although there is an easy to use
lexicon that let me tell it that www.TheHarrowGroup.com should have the
middle part pronounced as "The Harrow Group," rather than as an
understandably un-understandable mess.   But what Loquendo does that I
have not heard before, is that without any special tuning of any sort,
the resulting speech sounds vibrant, often quite natural, and -- well --
"human."  It's "expressive," in a way that I haven't previously heard
from computers.

But this isn't something that I can really explain -- it's something
that you really have to hear, so you can form your own opinion.  And you
have a couple of choices:

First, I've taken the text of this issue and had Loquendo's "Susan
(American English)" voice speak the issue into an MP3 file at this link:


http://www.theharrowgroup.com/articles/20020701/20020701-Loquendo.mp3 .


Although the good folks at Loquendo wanted to optimize the text to
assure that my spacing and punctuation didn't mess things up (following
a few rules, which I have not done, can improve things somewhat), I
wanted us to hear the result of a "typical" document that hadn't been
so-optimized, and that's what we have here.  The only significant thing
I've done is remove most of the URLs from the text, so that you don't
have to listen to lots of "double-you, double-you, double-yous..."  And
to me, the result is even more impressive than if I'd let them further
optimize it.

Second, Loquendo offers a demo site where you can type in a couple of
sentences in a language of your choice, pick the appropriate voice, and
it will translate and play the result for you, at
http://actor.loquendo.com/actordemo/default.asp?language=us .  You can
then find out more about Loquendo at
http://www.loquendo.com/en/demos/index.htm , or by contacting Gaea
Vilage at [log in to unmask]  . (If you do, tell her where you
heard about Loquendo.)

As you can tell, I'm impressed with Loquendo, and this is (of course)
just the beginning.  I can see the day when that old question, "Is it
Memorex, or is it real?" will take on a new meaning; at least in certain
contexts, we may have trouble deciding if we're listening to a computer,
or to a person.  And won't THAT be interesting...


VICUG-L is the Visually Impaired Computer User Group List.
To join or leave the list, send a message to
[log in to unmask]  In the body of the message, simply type
"subscribe vicug-l" or "unsubscribe vicug-l" without the quotations.
 VICUG-L is archived on the World Wide Web at
http://maelstrom.stjohns.edu/archives/vicug-l.html