I imagine it would take making a whole new resampler (and probably wavtool) designed to morph voicebanks together. With a plugin used to tell it which voicebank to morph. And then you'd just use flags to designate the intensity of the xsy.
The problem would be that it'd take a long time since...