Apple can’t even do realistic gesture tracking with their memojis on their highest end iOS devices. It’s appallingly bad and can pretty much only do a dozen or so expressions and with lag. I highly doubt their ability to do this in a more serious setting.

There’s enough issues in video conferencing today with audio codec compression and video buffering and Bluetooth issues - I would not want to add an entirely virtual layer on top of that. “Can you hear me?” will be replaced by “are you angry?” when it mistranslates a facial expression. Miscommunication over digital medium (text) is already a huge issue.