Most of the software that tries to isolate/remove a voice does so by differentiating between a voice [or voices] and other sounds or instruments, not by identifying a particular voice against another voice. Naturally if the speakers alternate then you could just separate them by manual editing, but it they overlap then of course that won't be possible.
Now, if it's a stereo file and the speakers are panned left & right that would be one way to try to separate them, or if one speaker is panned center and the other is off to the side then you could try using MS processing to isolate the mid or sides of the stereo signal. But otherwise I think you may be out of luck.
[I suppose you could try using Melodyne's DNA feature to see if it can isolate different pitches of two differently-pitched voices, but that's not really what's that's intended to do [it's meant to separate definitively-pitched musical notes, not voices], so I wouldn't really expect it to work]..
[There may be other software out there, but I'm not aware of anything that can isolate two overlapping voices]..