Night and day? No, more like distinct shades of gray. I'm approaching 70, with a slight ringing in my ears since someone tossed a fire cracker right next to me in an arena hallway at an Alice Cooper show 50 years ago. Filter settings make a significant difference in clarity of tone for me -- whether a reproduced guitar or piano string sounds distinctly real. Looking at graphs of the effects, I wonder if it has more do do with ringing and/or frequency alignment in time than their high frequency roll off differences -- which in theory I can't hear too.

Neither can you hear the ringing, which only happens at half sample rate frequency. Plus neither guitar, piano nor strings include any frequencies that could excite ringing.

Thanx, I deleted the false “default“ entries, as I have no overview about this.

SD Sharp is a good choice too that I like sound-wise.
Personally I tend to prefer the minimum phase filters about the linear phase.

I assume the pre-ringing of the linear phase filters is what I receive as smear.
Natural percussive instruments don’t have this, as there can be no sound before the drumstick (or whatever) excites the sound.

Trying to get the terminology right. Wondering if "ringing" has more than one meaning in different contexts. In standard graphing of DAC filtering, there are smaller waves before and/or after the main signal, implying the signal is smeared a bit temporally. I've seen that written up as "ringing." Not at all sure that these graphs are illustrating the difference that makes a difference to what I hear. There are some on the audiophile boards claiming that none of us can hear filter differences at all, folks preferring their technical theories saying we shouldn't to obvious experience that we do (although, granted, maybe they themselves don't). I'm curious to know the proper technical description of the source of the differences.

As a guitar collector, I have a decent sense of whether a reproduced guitar tone has realistic detail to it's attack and harmonics. What is it about a struck guitar or piano string (or cymbals for that matter) that make the realism of reproduction so sensitive to DAC filter implementation?

Thanks again. Just watched Monty's video. Enjoyed it; nicely done. Informative in regards to sine wave reproduction. Yet says nothing about the why you and I and others hear obvious differences between different filters.

A simple signal, like a recording of a flute or vibes, will come through pretty nicely on just about any system. Those are instruments producing notes relatively close to clean sine waves. Does the sine-wave-based analysis scale to complex musics, or is this like the three-body problem in astrophysics, where the complexity compounds far beyond what our current maths and analyses can handle, once more than a few simple sine waves (or two planetary bodies in the astrophysics) are involved? Intuition might tell us "Well, we have the math for two celestial bodies, so we can just use it for three." Or in this case "We have the math for N sine waves, so we can just use it for N+1." But math and logic don't always simply scale that way for complex systems.

I suggest not to create your own hypothesis’ or theories, but try to dig deeper into the existing ones.

Digital and analog audio signal transmission isn’t black magic, fortunately, but completely understood from the scientific perspective.

The two videos very competently cover some of the most important parts of digital audio, and demystify some points that are often misunderstood.
If you’re interested, watch them until you really get their informations.

Audio signal processing is deterministic and reproducable, you exactly know what comes out for a given input, and there is no unknown interaction between multiple parallel signals.
The other way round: the interactions, like aliasing noise and intermodulation distortions, are again predictable, for any number of audio objects, e.g. sine waves.
Music, of course, has a higher complexity than a few mixed sine waves, but still the result is scientifically fully predictable for any given moment of that ever changing signal.

Fortunately quality systems like ADI-2 are so close to perfect, that we are very close to in=out, in regards to what is audible to a human.
In=out is certainly true for sample rates of 88.2 kHz and up.

44.1 kHz and the resulting theoretical upper limit of 22.05 kHz for 1:1 digital sound transmission, and the filters that practically need to reach further down, are just close enough to our hearing capability that some changes to the sound become audible.

Regarding filters:
A filter (be it analog or digital) changes a signal, this is indeed it’s intended purpose.
That can come on cost of sound changes.
DA-filters typically are built to minimize audible side-effect of their job.

Minimize means this side-effects are not zero.
To judge the audible side-effect it’s easiest to listen to find your personal preference.
Theory and measurements can’t replace the listing, it’s the other way round:
Once you found you favorites ones, you can have a look at the measurements to find a correlation.

And– if you can’t hear a difference there is no relevant difference for you.

ADI-2/4 Pro SE BTW is great for such a filter and sample rate evaluation, as it can be feed directly with a fully analog signal from a vinyl disc.
ADI-2 Pro can do the same with a phono preamp fitted.

With all respect, that first video claimed that things are already perfect at the CD sample rate of 44.1. Yet we know from experience, and your next statement, that many of us do hear better musical reproduction at higher sample rates.

So our expert is clearly wrong about the science he's presenting, when he claims that 44.1 is already as perfect as we can possibly hear.

I'm an old man. The musical tonality difference I (and many others) hear is far below 20 kHz. The filter choices, on now three different ESS-based DACs I've used here, are audibly different to me. It's not just marginal, not just at the highest frequencies (which I can't hear anyway). Of the 7 stock ESS choices on their recent DACs, none sound quite right to me. Good vinyl with a high-end cart and stylus still provides better realism than 44.1 kHz signals through them.

The ADI-2's custom SD Slow filter finally produces a DAC circuit that's as pristine as the best vinyl pressings (assuming of course good engineering in the recording studio). Your team has pushed the art ahead of the science -- at least the science as represented in that first video, where the kid giving the presentation is so proud of his perfect theory that he persists in it despite real-life experience showing it to be, in the extended claims he makes for its perfection, incomplete. Denying evidence for the sake of preferring a theory, in its tight logic, has often held science back. As Thomas Kuhn's Structure of Scientific Revolutions showed, a generation schooled in particular doctrine often never surrenders it. It then takes a new generation to advance to better understanding.

Switching DAC filters causes audible tonal differences, not just near the high-frequency cutoff. There's surely doable science as to how. I'm on the side of science here -- and truly admire the ADI-2's scientific and artistic achievement. But I don't think we should accept the scientific project as complete until it explains what those of us with deep experience with listening to and playing music can clearly hear. Science still has work ahead of it. People thought physics was done in the 1890s too. It wasn't.

With complexly interacting components reproducing complex musical signals, we may be were we need to look at emergent features, in accord with complexity science: https://complexityexplained.github.io/. As George Box said, "All models are wrong, but some are useful."

This is not the case at all.

Did you conduct a proper double-blind testing? And even if you did, filters are characterized not only by their frequency response, but also by their phase response, which can lead to audible differences. No magic here.

There is no good vinyl. And please define "realism". If you claim that artifacts such as motor hum, detonation, stylus noise, clicks and pops make the reproduced sound closer to reality, then most will disagree.

OMG.

Typical audiophile rhetoric.

The case is much more complex than I thought initially.

Never claimed "magic" involved. Yes, I'm curious about the theory of how filters produce such obvious differences. Had time yesterday to watch the second recommended video, which brought in how filtering introduces aliasing. That's the sort of understanding I'm after.

The anti-audiophile response of "did you do a double-blind test" is unoriginal. The bias against audiophiles that predictably gets expressed in such cliches in online discussions is just weird. Yeah, I love music. Yeah, I can hear the differences in equipment. I also have some background in experimental psychology. Look up "change blindness." Obvious differences when two visual images are placed side-by-side are commonly missed when the two images are viewed in sequence. And we're got far more of our brains devoted to the visual than the aural. So doing A/B comparisons in the double-blind way is going to miss even large scale differences obvious to most everyone when comparisons are side-by-side. I've found, for instance, that in comparing speakers or amps, putting the same signal through two different ones at once make obvious differences which are not so apparent playing through them sequentially.

Virgin, unworn vinyl with a good cart and phono pre-amp. What I'm after, in particular, is whether guitars sound real. My comparison is with a lifetime's experience both attending shows and playing. I know what's realistic harmonics out of a well-attacked string. It's a matter of recognition. You can get there with digital and a good DAC, or with analog an a good cart, needle and pre-amp. As for recording engineers' preferences, there are those who work well with one set of equipment, obviously, and others who work well with another -- as with any instrument of artistic expression. Sturgeon's Law applies in this area too -- most engineers, as with most musicians, aren't that good. But the truly good ones -- beautiful stuff, done on very different sorts of instruments.

An audio comparison “side by side” would be a much stronger request and one step farther away than “blind A/B” from what I see at “audiophile” discussion, where comparisons are claimed with up to years between checks (“I had it but sold it”), different levels, different music etc. pp.

Practically I have no idea and would be interested how audio “side by side” can be executed in most cases, like filter comparison.

Regarding sound of natural instruments on recordings, from my point as recording engineer:
A recording is by no means the original, more like an as nice as possible photo, and the result is influenced by a dozen and more parameters.
Even the artist himself makes the same piece sound different on every take - intentionally or not.

You are right that the initial “attack” phase of any sound is the most important part for identifying it.

This is where a DA-filter can vary:
The form of the impulse, the first transient.
This happens in the high overtones only, but those largely influence how we subjectively receive the rest of the spectrum.

@Whit
Double blind test is the only way to get relevant results about object of study. Not only in acoustics, but many other fields.

Even side by side comparison has to be organized as double blind test to be relevant. If the differences are subtle and not clearly obvious.

Well put