I keep coming back to this forum post as the most relevant audiophile meme, in which someone finds an alternate memcpy implementation and claims to hear the difference:
https://www.audioasylum.com/forums/pcaudio/messages/11/11997...
https://www.audioasylum.com/forums/pcaudio/messages/11/11997...
“[..] this version removes the hard edge produced by memcpy”
“[..] most players use malloc to get memory while new is the c++ method and sounds better”
The same can often be said for todays onboard sound, no matter if via headphones, or active speakers. That's why some people are saying use the cheapest USB-sound-dongle, and enjoy decoupled sound, otherwise enjoy the buzz of your system-busses.
Also mainboard makers advertising some sort of shielding for the audio chip from the rest of the board, for the same reasons.
Memcopy produces a pattern on the memory-bus, which may be audible under some circumstances. Using a different pattern may "sound" different, or not all.
shrug