[orca-list] Possible bug with Orca and speech-dispatcher and selecting voices



Hello,

I have found several issues with orca, speech-dispatcher (abbreviated
spd in this message) or both when trying to use the mbrola voices
through the espeak-ng-mbrola-generic.conf spd file (using the generic
spd module).
1. In the Orca preferences GUI the Person label ends in (<lang>) <lang>
being the language code part of $LANG, regardless of the settings of spd
2. In the same drop down list, no other choice than the default is
available, even though several MBROLA voices are installed and checked
to work with espeak-ng.
3. These settings in /etc/speechd-dispatcher/speechd.conf:
DefaultVoiceType  "FEMALE1"
DefaultLanguage "fr"
should lead to usage of the fr4 mbrola voice, as in
/etc/speech-dispatcher/espeak-mbrola-generic.conf I have set:   
AddVoice        "fr"    "FEMALE1"     "fr4"
However, the language used by Orca is the one that corresponds to the
language part of the locale set in $LANG., e.g. "fr" for fr_FR.utf8,
"en" for en_US.utf8, at least if I keep in espeak-mbrola-generic.conf
the default command:
GenericExecuteSynth \
"espeak-ng -v mb-$VOICE -s $RATE -p $PITCH $PUNCT \'$DATA\'"
To get the fr4 voice I have to explicitly set it in this file:
GenericExecuteSynth \
"espeak-ng -v mb-$VOICE -s $RATE -p $PITCH $PUNCT \'$DATA\'"
But this is not an acceptable workaround as then this file would need to
be modified every time the user chooses another voice. 
4. The spd user settings written by spd-conf are no more effective than
the system ones.

Best regards,

Didier

On 07/16/2018 10:53 PM, Storm Dragon wrote:
Howdy all,

I have been working with the espeak-ng-mbrola-generic module for speech-dispatcher, trying to get it 
working with as many languages as possible. Some of the languages are very similar, e.g en-us or just plain 
en, pt-br vs play ol' pt, etc.

Everything works great for the most part. If I set the DefaultLanguageĀ  setting in speechd.conf to en-us, 
and the DefaultModule to espeak-ngmbrola-generic, then spd-say "hello world" uses the propper voice. Also, 
setting Fenrir to use the default module ifor speech-dispatcher uses the correct boice.

Orca, however, insists on using plain ol espeak-ng with the default en voice. It is set to use the default 
module (default synth), but something is getting passed through that causes the actual default to not be 
used.

Attached is the work in progress espeak-ng-mbrola-generic module, so if anyone is curious, you can see for 
yourself that it works in most places, just not with orca.

Thanks for any help with this,
Storm


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