Re: [g-a-devel] Happy patch bonanza

[Bill Haneman <Bill Haneman Sun COM>]

Hi Javier:

Actually, festival _is_ UTF-8 capable, at least for some voices.  The
change to use UTF-8 on the festival stream was part of a patch to
support an Indian dialect, but unfortunately it seems that the impact on
european languages wasn't understood when it was committed.  

I made a change to the gnome-speech festival driver in April 2005 to use
ISO-8859-1 as its encoding, which worked for all the festival european
languages which we were aware of at the time.  However it didn't work
with the Indian language above, thus the patch.

I still think ISO-8859-1 might be a better 'default' for the festival
driver than UTF-8, since as far as I know none of the european voices
expect UTF-8 input.  It works for English of course, but perhaps only
for ASCII characters!

regards

Bill

On Thu, 2006-06-29 at 07:57, "Dorado Mart�z, Francisco Javier" wrote:
> Hi all 
> 
> I would extend this solution to all speech-synthesis-drivers cause no TTS is
> UTF-8 capable. I have had this problem with dectalk and festival. And the
> other way, set to the locale We found that in Ubuntu for example, the locale
> is set to UTF-8 too. LANG='es_ES_UTF-8'	
> 
> Since some voices in Festival seems to work with UTF-8, I think a user
> setting to set the channel encoding would make sense.
> 
> Regards
> 
> Javier.
> 
> > -----Mensaje original-----
> > De: Bill Haneman [mailto:Bill Haneman Sun COM] 
> > Enviado el: jueves, 29 de junio de 2006 2:14
> > Para: Enrico Zini
> > CC: Hynek Hanke; William Walker Sun COM; 
> > zamazal brailcom org; ubuntu-accessibility lists ubuntu com; 
> > gnome-accessibility-devel gnome org
> > Asunto: Re: [g-a-devel] Happy patch bonanza
> > 
> > 
> > I think creating some configuration file, like for instance: 
> > gnome-speech/festival/voices.config
> > makes sense.  That way we can map voices to POSIX locales, 
> > which would be very useful, and we can also include info on 
> > the voice's string encoding format while we are at it.  It's 
> > a bit hacky but probably the best solution for now.
> > 
> > gnome-speech has some API for asking what locales a voice 
> > supports, but in order to implement it better such a config 
> > file is needed anyway. It might as well be a simple text 
> > format so that users can easily add voices.
> > 
> > Bill
> > 
> > 
> > On Wed, 2006-06-28 at 23:48, Enrico Zini wrote:
> > > On Wed, Jun 28, 2006 at 09:22:41PM +0200, Hynek Hanke wrote:
> > > 
> > > > if you use festival-freebsoft-utils to communicate with Festival, 
> > > > then you can send all the input in UTF-8 through the appropriate 
> > > > functions and let Festival care about the necessary conversions 
> > > > between encodings. Encodings can be easily defined by the user in 
> > > > the configuration file, or can be specified by the author of the 
> > > > voice, as is the case with festival-czech. It has a dependency on 
> > > > the 'recode' utility.
> > > 
> > > The gnome-speech festival driver just runs "festival 
> > -server" and then 
> > > communicates with it on port 1314.  I don't know how much effort it 
> > > will be to convert it to use festival-freebsoft-utils, also because 
> > > there seems to be a general consensus in moving away from 
> > > gnome-speech.  From what I understand it's currently fine to make 
> > > fixes to gnome-speech, but a bit too late to do major redesigns.
> > > 
> > > BTW, I now realise that by the time the Italian Festival voice can 
> > > understand UTF-8, we'll definitely have moved away from 
> > gnome-speech, 
> > > and since a long time, too.
> > > 
> > > 
> > > > If you want to go some other way, I'd highly recommend that the 
> > > > encoding used for different voices is easily configurable by the 
> > > > user. I think there is no way how to determine the encoding of a 
> > > > given voice in Festival automatically (which is of course 
> > broken :( 
> > > > ), so giving the user the power to fix the problem without 
> > > > recompiling anything is very important.
> > > 
> > > Good point.  So, either there's a way to query the 
> > preferred encoding 
> > > to the festival voice, then we should use it.  Otherwise, it should 
> > > all be read from some external config and not compiled in.
> > > 
> > > There's also a way halfway through, that is adding to the Italian 
> > > speech synthesis LISP commands to let gnome-speech know of the 
> > > encoding, or to do the transcoding.  I unfortunately don't 
> > know enough 
> > > of Festival to be able to do that.  Maybe the festival 
> > developers can 
> > > help here?
> > > 
> > > 
> > > Ciao,
> > > 
> > > Enrico
> > > 
> > > --
> > > GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini <enrico debian org>
> > > 
> > > 
> > ______________________________________________________________________
> > > _______________________________________________
> > > Gnome-accessibility-devel mailing list 
> > > Gnome-accessibility-devel gnome org
> > > http://mail.gnome.org/mailman/listinfo/gnome-accessibility-devel
> > 
> > _______________________________________________
> > Gnome-accessibility-devel mailing list 
> > Gnome-accessibility-devel gnome org
> > http://mail.gnome.org/mailman/listinfo/gnome-accessibility-devel
> > 
> _______________________________________________
> Gnome-accessibility-devel mailing list
> Gnome-accessibility-devel gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-accessibility-devel