I don't hear it as exclusively on s and sh sounds, more global in mostly upper mids. I've long found upper mids is where I almost always first hear any tonality, timbre defects. Thinking about it a bit more, could also be described as a bit of nervousness. In spite of increased perception of performers in room, therefore, even more dimensional imaging, transparency, lowest level performers, instruments greater presence (this added presence is the biggie of the newly added network switch). I would say I'm now at extreme level of resolution, transparency, every single thing in streaming, digital setup optimized over many years of analysis and listening. Can't think of a single thing needing more optimization, only wholesale changes like new server, dac would change the equation.
The question becomes is the 005 to blame, delta sigma has had the reputation of being least analog sounding dac topology. Am I running up against the limitations of DS chip in 005? Or is this coloration somewhere else in system? At this point I'm going to assume something else in system. I do have a fair bit of rhodium and silver in system at this point, can change out some for gold. I still have many other options as well at this point, so not going to jump to conclusion 005 is culprit.
I'm also going to need to listen more closely if there is a difference between streaming and my cd rips. If there is difference could also point to music player software. I'm presently running Roon with Tidal and Qobuz. Formerly ran Audirvana some years ago, thought I perceived a slightly less analog sound when first changing to Roon. Loved the interface and thought sound quality close enough so kept Roon. May have to retry Audirvana, looks like they have new interface with both Tidal and Qobuz. Haven't searched for online comparisons lately, but recall some believed Audirvana sound quality superior to Roon a few years ago.
I'll continue to say streaming is the most confounding and complex thing I've ever encountered in audio, so many directions one can go in, so many choices. At level of full optimization, unique setups make any generalizations problematic.