its another thing to have
A. not perfect impulse response because of "naturally" occouring errors, we are used to from our day to day life
B. have a NOT naturally impulse response that basicly affects "consistently in a bad way" a good portion of the overall sound/frequency spectrum
you are free to see this differently but it just makes sense, atleast for me...
tho for me also absolute phase (specially under 1-2khz) is crucial and others complain it cant be heared xD it is what it is i guess
sure errors exist, but you can be 10 degree off OR 190 degree....
C. Continuing to listen with our eyes and misinterpret visuals from impulse responses to figure out how thing sound. Usually the most intuitive false conclusion becomes that no filter or very weak filtering increases accuracy because the response
looks more like a Dirac pulse.
I’m so fed up with those intuitive BS conclusions from impulse responses.
Ringing is the mathematically accurate result of a brickwalled signal. Which is what we want for accurate reconstruction of already band limited digital recording. Treating the signal as a sum of sines, and the impulse as an infinite sum of sines, the best band limiting will look like ringing. It’s not a fault, and it certainly isn’t how music sounds.
If you use some massive EQ in the bass with a filter that gives pre ringing, then you can end up hearing a sort of pre echo before the impact of the bass. That sure is weird and undesirable.
But repeat the same at higher frequencies and the audible impact rapidly becomes smaller and smaller. To me it stops being noticeable well before I reach the treble.
If I was paranoid about that anyway, because it’s the one sound I can actuallly correlate with impulse ringing, I guess I would get a DAC filter that is minimum phase and be done. But I don’t even manage to care because audibility falls down so much, plus the music content is tens of dB below FS even before filtering, and my own sensitivity add another tens of dB at those freqs. I can’t convince myself to care.
Now if you use various filters on a DAC, several might be audibly different(and not just because your eyes tell you to hear a difference like you usually do). But AFAIK, it’s because they allow obvious crap as optional setting. Either the amplitude difference is at frequencies we can still hear, or an absurd amount of crap aliased back into the audible range, or in case some high energy ultrasounds remains, sometimes it could be IMD or some amp oscillating like mad.
Proving the audibility of bad choices was never hard. Does it mean we should care about filters? Is the main reason to care, the fact that we get to select between many crap settings beside one or 2 common sense settings? It’s so inconsequential that many people still manage to convince themselves that inferior fidelity is higher fidelity.
The most typical filters turn out to be the ones that avoid those bad side effects.It’s almost as if engineers had figured it out before audiophile marketing started to try reinventing the wheel...
Of course engineers are also focused on stuff like money and computational power, which also tend to lead to close enough results whith conservative solutions, because while you can throw money at this and use more processing than crypto, the extra improvement follows a strong diminishing return where audibility rapidly gets unlikely, then impossible(and I sure don’t enjoy waiting 3 seconds anytime I press "play" or "next" on my player anyway).
About apodizing/windowing, I remember some paper suggesting it could be interesting in the studios. I think there was some reason to discourage double use. It was a while back and as usual the math rapidly went over my head. It was already with some of the Meridian people so IDK what it’s worth(did they coin apodizing for audio?). Maybe they had not yet considered making way more money from selling that stuff in their DACs. It’s frustrating how many clever and skilled people they had and worked with, and how often they still decided to BS their way into marketing.