Speakers output back as input node

Hello,

I'd like to ask if it's possible to get the speakers output and send as
input with web audio API. The idea is to stream (or send by PeerConnection)
anything that one listens too - e.g. music tracks from MP3 player. So
instead of sending out microphone input by WebRTC I'd like to send speakers
output.

I've noticed Web Audio is working on some integration with WebRTC, so I'm
wondering if there is a chance if this could be implemented if it's not
offered by the API already.

Best wishes,
Marek Tłuczek