To get vocal/formant sounds out of a DX100, you would need 2 things (which I'm not sure it has): 1)the ability to assign fixed frequency to some of the ops; and 2) an algorithm that branches down treelike, 3 to 1. Within a patch you can set sensitivity of some modulators but most of that control is global in the FUNCTION mode.
If you have trouble finding specific advice and patchbanks, you can email me and I can send you diagrams of DX7 patches that illustrate what I'm talking about. You would have to ignore or compensate for the 2 extra ops but they would still work.

mikenels@gateway.net
good luck.mn