Does the Emic 2 have support for loading a custom voice in either SAPI or in any other format? If so, can you provide a pointer to how one would load a custom voice.
Emic2 was designed by Joe Grand, all of the documentation regarding this module can be found here: http://www.grandideastudio.com/emic-2-text-to-speech-module/ .
It appears that there is the ability while in the DECtalk parser mode (using :dv design voice) to change a lot of a voice characteristics, there are some examples in the Parallax Text-To-Speech PDF.
The DECtalk documentation references SAPI, but doesn't go into specifics about SAPI.
I don't find or know of any "loading" of custom voices into Emic2, there only seems to be the ability for you to create your own text sequences containing the ":dv" command with with DECtalk parameters to change voice characteristics.
On Joe's site it is possible to contact him and perhaps he could elaborate/help you with your question concerning loading.
I just saw Joe last weekend in the Hardware Hacking Village at DEF CON, so I know he's alive and well.