I slurped the existing dialog-handling stuff* into speech.cpp, and added a field to allow differntiating speakers, an entry looks like this now:
{
"type" : "speech",
"speaker" : "migo",
"sound" : "\"Hello?\"",
"volume" : 10
},
{
"type" : "speech",
"speaker" : "creepy_doll",
"sound" : "Do you really need that much honey?",
"volume" : 10
}
It’d be pretty easy to add some optional fields for mood or similar.
If we want to do it with more monsters, we probably want to dedicate a speech flag to it that matches the “speaker” field I added, that way you can add everything you need in json. I can see a mod with zombies walking around saying random crap being really funny, just not for the main game 
*Completely independent of this request, heh.