This is an interesting discussion to me because I used to design and build loudspeakers with time alignment and phase-coherent filters, and I cannot say I hear the same effects across the board using the upsampling capabilities of my dCS Upsamplers, and I’m mainly a classical music buff as well, with the usual (anyone in the know, it’s kind of sad…) preference for old recordings, as well as some jazz and rock music. There are recording in which upsampling seems to make sense of spatial cues that are audible as an almost separate entity without, and there are recordings that appear slightly blurred or bloated or phasey with upsampling, in other words, I’m not in principle in favor or against using it, and in case of doubt, especially of course when it comes to modern recordings, tend to leave it on. Also, it very much depends on the native recording format apart from microphone placement etc. I love one-point recordings, but even there, there’s no rule that upsampling will affect the phase response in a negative way. It may have more to do with the speakers and setup one uses than the actual recording. I’m always surprised how many so-called reference speakers e.g. use midrange drivers in inverted polarity etc., produce dips in the spectrum, use hard cone materials that break up and ring at higher frequencies, not to mention the lack of room treatment in most audiophiles’ homes, so my best advice would be to go by the individual recording or album.
Greetings from Switzerland, David.