I found this pretty interesting - not an audio application but the principle is the same. Measurements are on a CAN control bus in an grid connected inverter. Heaps of noise on the signal lines (that's another problem - still being investigated). Sorry about the measurement quality, a 10 MHz 8 pit Picoscope was all I could fit in the overloaded toolbox and still be allowed on the plane. Anyway, although the CAN Low and CAN High were unusable, the differential on the twisted pair had most of the common mode noise cancelled out (purple trace at the bottom) was good enough for the application. Always nice to see the theory play out in practice.