Tons of hacks exist in various avatars to try and "approximate" the visemes that are provided by vrchat's viseme blendshape feature. This is required for many purposes:
* The wanted viseme actions involve bone rotations or other advanced behaviour
* The user wants viseme blendshapes on multiple meshes
Please expose the current level of each viseme as a separate avatar parameter (ex. Viseme_PP Viseme_FF etc) so that we can eliminate these poorly-performing and inaccurate approximations based on Voice and Viseme.