Is this why high sample rate music files might sound differentially better on DACs with less effectively implemented filters, while red book and even mp3/4 or ACC files can sound pretty great on DACs with more sophisticated/better implementation of filtering?
First you'd need to demonstrate the audible difference you are describing actually exists - rather than being bias based.
OK, maybe it’s only gopher s-h-i-t. I have at least 100 titles on vinyl that were recorded, produced and pressed before 1978, and very likely have zero digital elements in the recording chain. Some of my other boutique titles are much newer and claim to be recorded using 100% analog/tube gear from the microphone to vinyl, and I also own some MoFi reissues that I presumed to be 100% analog but turn out to have a digital step their production (they sound glorious in spite or because of this).
I was ignoring dithering, that's true. The linked site which was bein discussed also does so, as far as I can see. So with that, the error will be zero on average and uncorelated but not zero for individual samples, right?
Those Mofi digital steps are probably DSD masters and are going to be identical to the analog tapes they were archived from. I don't have an issue with digital recording or mastering as long as it isn't highly compressed. With some modern records the most important part about them saying all analog is they aren't compressing the audio during mastering. There are some amazing sounding digital albums though. Herb Alpert's Rise and Donald Fagan's Nightfly were recorded on 3M PCM digital equipment from the late 70s and those albums sounded amazing on the original vinyl records or the later 80's CDs.
Aren't such low frequencies considered to be non-localizable?
Time difference between what 2 things?
You can check the sources, various stimuli were tested: Gaps in playback, double-impulses, time differences between the ears for the same sound and so on. All with different outcomes. That's also why it is difficult to determine which stimulus would best apply in a case like this where some small shift in real music is being proposed.
If tmin in the equation in the link provided by @RandomEar earlier in this thread is demonstrably related to our current understanding of the lower limit of the sensitivity of human hearing for timing errors, then the answer is “yes”. For low frequencies and low amplitude signals, tmin approaches 10us which is near the limit of human hearing to differentiate timing errors, so signals in this range are certainly not orders of magnitude below our ability to detect timing differences.
As pointed out by @popej , the calculation I linked and my own example disregarded dithering, which is standard in music mastering. So the actual error is (on average) way smaller than that theoretical boundary without dithering.
Of course, which is why LFE for home theater is provided in mono and generally no one complains. Nevertheless, bass below 80Hz in stereo recordings “sounds better” to me and other listeners on some recordings played back through some systems and in some formats. More clarity, appropriate decay, etc are clearly discernible as “better”.
On this point we agree completely.
Is this really because of the stereo, or it’s more about the dual subwoofer sources—even dual-mono—and their known benefit in averaging the room peaks & dips?
That's only true when looking at time-of-flight difference-based localization cues. Level differences are fully effective at all frequencies (as can be easily demonstrated with headphones).
But level differences become smaller and smaller with decreasing frequency, because the dampening effect of the head is reduced. So for any "real" sources (as in: not artificially created level differences by using headphones or DSP tricks using speaker arrays and such), this effect should become equally insufficient for localisation.
