[PATCH 4/7] replace EVoiceType with SPDVoiceType

[Andrei Kholodnyi]

> good, I'll provide a patch today then.

it will take an another day.

There is a small problem...
it is possible to set in the module  either voice or synthesis_voice or both.
now what setting has a preference?
At the moment it is resolved in a way that in each synth module we have e.g.
        UPDATE_PARAMETER(voice, espeak_set_voice);
        UPDATE_STRING_PARAMETER(synthesis_voice, espeak_set_synthesis_voice);

where implicitly synthesis_voice setting has preference over voice,
but not always :)
it will be updated only if values between two set operations are different.

E.g. if I do set
male1, bosnian
male2, bosnian

then at the end male2 will be set

However if I do set

male1, bosnian
male2, english

then at the end english will be set

Moreover if I decide to swap in the module UPDATE statements

        UPDATE_STRING_PARAMETER(synthesis_voice, espeak_set_synthesis_voice);
        UPDATE_PARAMETER(voice, espeak_set_voice);

then behavior will be different and not consistent with other synths.

It looks like we need to move UPDATE statements to the module_utils
and not rely on synth implementation.

Another point of how we want to handle a situation when we have both
voice and synth_voice set?
Do we what that synth_voice has a preference over voice settings?
I think this is a behavior now and we want probably stick with it

Andrei.