Allow actually setting the synthesis voice when then generic module is used

[Didier Spaier]

Additionally, with this patch applied, Chromevox in Chromium speaks the
espeak-ng-mbrola-generic voices.

So the issue was not in Chromium, after all. ;)

Best,

Didier 

On 10/11/18 12:33 AM, Didier Spaier wrote:
> Hello,
> 
> On 10/10/18 1:50 AM, Samuel Thibault wrote:
>> Hello,
>>
>> Didier Spaier, le mer. 10 oct. 2018 00:41:56 +0200, a ecrit:
>>> +char *module_setvoice(char *voicename)
>>> +{
>>> +   int i;
>>> +   for (i = 0; generic_voices_list[i] != NULL; i++) {
>>> +   if (strcasecmp(generic_voices[i].name, voicename) == 0)
>>> +           return voicename;
>>> +   }
>>> +   return NULL;
>>> +}
>>
>> It seems odd to call this function module_setvoice since it does not
>> actually set anything, it just tests the presence in the list. Perhaps
>> rather call it module_existsvoice and make it return a gboolean?
>>
>> Samuel
> 
> The updated attached patch addresses this concern.
> 
> Logic implemented:
> 
> If the voice name is set by the client the language and voice types are
> not considered.
> if the voice name is not set by the client it is deducted from the
> language and the voice type, set as default values or by the client,
> as previously.
> 
> This implies that if the voice name voice is set by the client, the
> language and voice type (set as default values or by the client) can
> differ from those associated to the voice name in the configuration
> file, but I don't think that matters: if the voice name is set by the
> client it should have priority over what the AddVoice line says.
> 
> Didier
> 
> 
> _______________________________________________
> Speechd mailing list
> Speechd at lists.freebsoft.org
> http://lists.freebsoft.org/mailman/listinfo/speechd
>