In the linked example it's a private voice room. You an use that one browser source to change the avatars of everyone in the voice call, via the UserID targeting. This guy is only running _one_ browser source and using the discord 'detect speaking' to change his avatar. (And if doing one browser source per user, then you probably don't even need to do the Discord Dev mode/copy ID stuff, you can just target on `.speaking` instead, this guy did it in a convoluted way for his use case)
And with streamkit setup like this it doesn't lend itself to having multiple people in the same voice room. Unless you are getting _everyone_ in the voice call to boot up into different voice channels and sending their streamkit URL to you, which would be silly, as then you can't even hear each other as you are in different rooms. That doesn't work how you think it works, you don't need multiple browser sources.
