Not even sure to what extent it's possible, but this is what the concept is all about: being to emit messages simultaneously just like in real conversation, but being able to receive them sequentially. Sending the transcript is already cool, but if we could send the audio simultaneously (in a way that allows the user to play it back), that'd be even more awesome...
If someone comes around here and knows getUserMedia well before I leap into it, I'd love to talk with you.