I realize CT 5.0 may have some changes in this regard, but if not (or even if so :>) one thing that could improve CT would be to have a randomly seeded head and eye blink movement for every animation.

As it stands now, in order to generate multiple characters using multiple passes, you end up with all of your characters with the exact same shoulder/head and eye blinks, as if they are coordinating synchonized dancers (this is true even if you put eye and head movements in the timeline -- during the times in between movements the default behaviors are the same).

There is a workaround -- you can insert silence into your .WAV file so the two characters lines aren't the same length and then trim in post, but this is a PITA (especially if you forget to do this initially).