I don't. One of the reasons for getting the X3ii was to figure out this debate, at least on reasonably priced hardware. On my big optical spinner, I can discern the SACD layer from the CD layer on some tracks (but I don't know whether this is due to resolution or to the way the CD data was cooked). On the X3ii (with the same amplifier and headphones), comparing carefully chosen short samples, I cannot hear any difference between 192kHz 24-bit and the 48kHz 16-bit versions I downsampled from them. By the way, I learned that lowering the sample rate can be done in many ways, and some of these lead to changes that I think I can hear, so I first focused on the simplest approach, called downsampling in sox: keep every Nth sample as is, completely ignore the N-1 samples in between, and obviously play them at a slower pace (according to the documentation, this could lead to "aliasing", but as said, I can't hear any difference). Even at 32 kHz, throwing out 5/6 of the samples, it's quite a subtle difference to hunt down. At 24kHz, it's obvious when comparing, but I would probably never realize otherwise. At 16kHz it's glaringly obvious, like a mono radio switched from FM to AM.