There is no convincing evidence that a Mozart symphony has actually been transmitted faster than 'c'.
Suppose you have a psychic, who claims to receive messages faster than light.
You decide to put her to the test. So you send her the following sequence:
1,2,3,4,5,6,7,8,9
and ask her what the next number is. She correctly predicts 10, and does so at such a time that she couldn't have recived the number '10' by any possible light signal.
Are you convinced that she received psychic messages "faster than light"? I'm not.
The problem is that predicting the next element of the sequence {1,2,3,4,5,6,7,8,9,x} is not "hard" - you can do it by linear interpolation.
Draw a waveform of a Mozart symphony, record it, and plot the value of the signal at 1 nanosecond intervals. You will see that it is essentially a straight line. Thus sending a Mozart symphony is not a good test for whether one has sent a signal unless one sends it quite a long distance. It's no harder to predict the value of a Mozart symphony correctly 1 ns in advance than it is to predict that 10 was the next number in the above sequence. In both cases, one uses simple linear interpolation.
To provide convincing evidence, you'd have to send a random sequence, not an easily predicted one. A Mozart symphony is simply not random, and not a good test.
The technique in question fails the test if one attempts to send random information (such as white noise) rather than a Mozart symphony.