Thanks - that is very clear. I’ll keep SRC off unless I need it.

The upsampling chip uses it‘s own anti-aliasing digital filter on the incoming sample rate.
So with upsampling you can‘t avoid DA filtering in the audible band.

When running in native sample rate you have the choice to select the ADI-2 DA-filter that suits the music and your taste.

To me too, the SRC sounds quite good, on par with the DA-filters, and better then the standard “Sharp” filter that I dislike, kind of.

If you like to go one step further, try offline upsampling.
As offline doesn‘t have any time/latency constraints, it can use much longer filter-coefficients, which can lead to even better results.

The differences are subtle, but nevertheless...

A lot of so called Hi-Res stuff in streaming is simply upsampled, visible if the spectrum content doesn’t reach above ~22 kHz.

I do use 192 kHz for analog feed-through too, like vinyl on the ADI-2/4 Pro SE.
The high SR has no negative aspects, so why not.

For offline resampling (= creation of a new audio file) I’d say 88.2 kHz / 24 bit is sufficient for the full positive effect.
This is just 3 times CD file size, manageable with storage these days.
192 kHz / 32 bit is almost 9 times CD, so ...

Realtime resampling, like the built in hardware SRC, is comparable to what the DA chip does internally anyway, just another filter color more or less.
If you like the sound, go for it, but even consider selecting the DA-Filter that suits your taste, instead.

Once I’m above 48 kHz the DA-Filter doesn’t matter any more, at least I can’t discern any audible differences between the filters.
That’s why I think higher than 88.2 kHz doesn’t improve anything more.

But not to forget, to up the SR there‘s always a filter on the base sample rate in the game, so don‘t expect wonders.