I've mentioned this before, elsewhere - but being able to create, manipulate and delete visemes without importing an audio track, or text to speech input would allow "cleaner" facial animation. There would not be a need to spend so much time cleaning up the viseme tracks as is the current workflow. This would allow us to enhance the current range of expressions by fine tuning lip, tongue and jaw movement quicker and easier than we can currently do with set frame, puppeteering and auto-lipsync options.