That is, you normally have a Head layer, under that a Mouth layer, under that the visemes. These layers are needed. I don't think you have a Mouth layer for example, so the lip sync cannot find the visemes (it looks at the children of Mouth). The Head layer I think is needed for the Face behavior to use the webcam for Smile and Surprised visemes.
So try restructuring the layers and see if that fixes it.