[BjoernKW]

The codecs and sampling rates used for phone communication are optimised for the usual range (E2 to C4 depending on voice type) and overtones of the human speaking voice. The usual sampling rate for phone lines is 8 kHz, which means that only frequencies up to 4 kHz are captured. This is sufficient for making vocal communication intelligible (if not exactly aesthetically pleasing).

Music typically goes beyond that range. When transferring music over a phone line overtones in particular are cut off, which makes the music sound flat and creaky.