2nd: Comments on the Text to Speech "algorithm"

[Storm Dragon]

Hi,
It might be interesting to make voices where the sounds of the vowels
are recorded and the harder sounds are computer generated. This would
have far fewer recordings than trying to record every word, may be much
faster, and might sound close to human speech. Of course, it may also
fail miserably, but it may be worth a try.
Storm


-- 
Follow me on Twitter:
http://www.twitter.com/stormdragon2976
My blog, Thoughts of a Dragon:
http://www.stormdragon.us/
What color dragon are you?
http://quizfarm.com/quizzes/new/alustriel07/what-color-dragon-would-you-be/



On Fri, 2010-03-05 at 12:44 -0600, William Hubbs wrote:

> Hi Marc,
> 
> On Fri, Mar 05, 2010 at 01:19:59PM +0100, marc wrote:
> > Well,
> > 
> > Now I've looked into vinux, and I must say: quite fast voice, nearly 
> > unreadable to me.
> 
> I use my speech synthesizers at around 100 wpm, and some would say that
> is slow.  I am not convinced that an idea like what you are talking
> about would be able to perform at high speed.  Not to mention, you would
> have to record every possible word in each language you want to support,
> and if you are supporting tonal languages (languages that assign
> different meanings based on the inflections of the words as well), you
> have to take all of the possible inflections into account as well.  So,
> you are talking about a very large database of recordings.
> 
> I just don't see how an idea like this would come close to being
> practical.
> 
> William
> 
> _______________________________________________
> Speechd mailing list
> Speechd at lists.freebsoft.org
> http://lists.freebsoft.org/mailman/listinfo/speechd
-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<http://lists.freebsoft.org/pipermail/speechd/attachments/20100305/e2460011/attachment.htm>