When applying a file created on TTS on demand, the viseme timeline needs to be edited to include a "none" at the end so as to not mess up ensuing expressions.

In the example I've sent I show a file created with tts on demand. Without editing the adding smile is distorted. The second half of the example shows what it looks like when the file is edited.
Example link:
https://www.dropbox.com/s/i2oq5pyr0cag5cv/Example_1.wmv?dl=0

It would be helpful if the TTS file could be generated with a "None" at the end.