The scale of your model is too small for the faces to form correctly.
Scale the model up and the down again when complete or use the Dave Method as described here.
I get a few comments about the lack of actual sound, but with a face made for radio and a voice made for silent movies…that’s why I use ‘subtitles’. Scaling (as mentioned by Box) is your saviour with these things. Model at 10 or 100 x the desired size, scale the final product back down.