I have a small suggestion: add a toggable, adjustable "hold" delay for the open-mouth frame. The idea would be that with it activated, it stays in that image if sound comes into your mic before the set delay time passes.
I'm using images that are part of an animation, and with your software, it looks like my character is, let's say, having a VERY bad time, haha! And I think a solution like I suggested will fix it pretty much completely. That's the only think keeping me from using your solution for good instead of my current one.
Thanks, and keep up the excellent work!