Re: [orca-list] eSpeak, Speech-dispatcher, Orca and voice variations in Gutsy

I have two parts to discuss here.

Firstly, I have speech-dispatcher 0.6.4 and espeak 1.29 on a slackware
12.0 system. I am able to get a female voice by changing the default
voice in the speechd.conf file to something like FEMALE1. The question
was how to get Orca to use this voice, as it was working with spd-say.
The voice you would need to use in the orca settings is espeak default
(not default). Whilst writing this email I have a female voice from
espeak coming from orca.

Now to the other point. Tomas said that what should be done to make the
voice varients available. As it is planned to have orca using language
selection and then choose a voice for the language, I would say, that
can wait until it is planned (if something could happen to
speech-dispatcher). Tomas was suggesting having a way of exposing the
varients you want in the espeak.conf file, well this would be good, but
why not just have it that the voices mapped to MALE1, MALE2, ...,
FEMALE1, FEMALE2, etc could be defined in espeak.conf as the default
voice could be set in speechd.conf and then used by orca. Not perfect,
but it might be enough for the mean time.

Michael Whapples
On Thu, 2007-11-22 at 13:56 +0100, Tomas Cerha wrote:
Jonathan Duddington wrote:
  espeak -v en+12 "hello world"

Male and female variants are:
  m1  m2  m3  m4  m5   f1  f2  f3  f4

Thanks Jonathan for explanation.  Thus I guess we'd need to offer all
these variants for each voice in the SD output module.  The problem
might be the huge total number of voices in the selection.

This could be solved nicely if Orca first allowed you to select the
language, so that the voice selection would then only contain the
available voices for one language.

As I talked about that with Will on this years Guadec, this is planed.
Thus the question is whether we better wait for this change in Orca or
whether we decide this is so important, that we want it now regardless
the huge voice selection control.

Another option might be to allow custom voice definitions in espeak
module configuration, which would map symbolic voices (exposed to Orca
configuration) to real espeak voices.  This might be also an interesting

Best regards


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]