Here's another posibillity: Create a wave file with the entire sound recording of a conversation, and then create the mimic file. It will have only one of the 2 characters having the entire conversation. But if you cut out the key frames for the parts that character B is supposed to talk then character A is set - he'll only be talking for his part of the conversation. Re-apply the mimic pose to character B. Cut out all of character A's speech keyframes (keeping the ones you cut out of character A's) and then it should look like they are having a conversation. I haven't tried this yet, but it seem like it should work. Any takers?