Most of my music listening is on speakers, not headphones, but for really critical things like ABX testing, headphones simply reveal more in midrange and high frequencies. I mostly listen to jazz, pop, rock and classical. When I compare lossy vs. lossless and various sampling rates and bit depths with headphones, I have a problem: the masters of the test tracks typically presented, usually pop and rock selections, mostly sound so crappy that it hardly matters what the resolution of the distribution medium is. I mixed broadcast TV production audio for 30 years and I find spoken word or a solo singer very useful as test material because (a) I’m familiar with real sound of the source, and (b) the stark contrast between the words and a dead quiet background leave no place for audio reproduction artifacts to hide. Getting the original recording right is key. Unfortunately, it’s seldom achieved in the music industry. (The movie industry and BBC radio would make better sources of suitable test material for me.) When I take an ABX test it’s usually: “the vocals are distorted. Next…” over and over again.