When I worked with boardroom tele-conferencing equipment, it used DSP as a digital echo canceller to make a model of the reflections and phases in a room to stop sounds from the speakers from being transmitted back to the other end where they came from.
It tried to make "full duplex" but frequently cheated by using some attenuation when the equipment was receiving and transmitting simultaneously.
The DSP varied the model of the room slightly when a door opened or when somebody moved.
Maybe DSP could be used here to cancel the music that is picked up by the microphone so that it could sense noises.