That's not how audio codecs encode audio though. If we directly cut off bands that were difficult to encode, most codecs would stop writing anything above 1500hz, and then you'd be listening to a useless mumbled mess. A lossy audio codec will use the FFT algorithm to encode the entire spectrum, and the quality controls how aggressive the algorithm quantizes the waveform (through a process called sampling). On a lossless file, the compression will decode to an identical representation of the original master.

>If we directly cut off bands that were difficult to encode

not difficult to encode, inaudible to humans

>entire spectrum

oh boy, so why does Hi-Res stop at 192KHz? clearly its not the entire spectrum. We should be encoding at least to couple MHz, right? more better!

Hehe. You jest but MHz is the range of ultrasound and is pretty useless in air.