The switching frequency guidocorona, may be 520khz, but it’s the amplitude as well of it not just the frequency and has to be filtered at the end with all that rms power by low order filters that have to take that power.
From what I’ve been told by those in the know of Class D technology this has to be shifted up 5 to 10 times higher so the filtering of it on the output of a Class D amp can then also be shifted up higher, and therefore not be affecting the audio band, with audio band HF phase shifts or audio band HF roll-offs.
On most of todays Class D amps that try to get the highest frequency response they can get, you may hear a sizzle from a high bandwidth tweeter if you put your ear to it, if it’s not muted eg: in-between tracks on a cd.
Then there are those Class D’s that want to get rid of that sizzle which filter it even more savagely, then those ones can be starting to roll off the audio HF already at 5khz.
The sounds of the two can be then different in the upper-mids and highs, one being hard or harsh because it lets through some of the switching noise. The other sounding opaque or soft in the upper mids and highs.
Therefore to cure one or the other the switching frequency needs to be much higher so the filtering has little or no effect within the audio band.
Analogy: is to remember the early days of cdp’s with their brick wall filters, which on paper looked ok but sounded shocking, just because they wanted to show a flat 20hz-20khz frequency response. But in Class D amps this filtering is low order and happens at the end of the power amp at full rms wattage power levels, and the filters need to take all that power handling.
Then there's the SMP power supplies another problem again, that some Class D manufactures try to get around by powering with more expensive to make Linear power supplies.
Cheers George