A memory device with no moving parts has been an objective for many years, and will soon be practical even for very large blocks of data like the digital representation of high bandwidth music. This is separate from the question of how the digital representation is formatted. A static memory device could output a data stream exactly like that obtained from a spinning disc.
I have no significant experience with MP3, but I gather that this protocol has used data compression as a way to put more music into less data space, with "acceptable" fidelity. However, don't condemn all data compression. It can also be used to provide greater resolution from an available data space. Why leave the highest order bits of your data as zeros throughout most of the music, just so you can make them ones for a few seconds? With a few exceptions (specialized computers in military equipment) computers use Floating Point data format so that they can achieve great accuracy using a reasonable number of bits in a word. Digital music should do this also.
