would someone be willing to provide two such testfiles? without telling which file is which and i post my results?
Resampling uses the same types of filters with the same filter requirements and obviously there are many different individual filter designs/variations. So “yes”, the different filters do change the “sound signature” (FR, phase and/or “ringing characteristics”) but NOT audibly so in most cases. It’s only in pathological cases where there’s an audible difference and such filters can justifiably be considered “faulty”, as already mentioned!
So you’re saying that because we’re “still happy with it” the EQ plugin we’ve apparently subconsciously inserted and adjusted (to compensate for filter “flavour”) is invisible/can’t be seen? That doesn’t make any sense, how does a plugin become invisible and how would we adjust it (even subconsciously) if it were invisible? We’re “still happy with it”, even after numerous resampling applications, because it’s not audible and there is nothing to compensate for!
And which DACs allow you to disable oversampling? All DACs resample/oversample (with the exception of NOS DACs). If you are disabling oversampling/resampling in say CamillaDSP, you are NOT disabling resampling/oversampling, your DAC is still oversampling (assuming you’re not using a silly NOS DAC)! What you are disabling (in say CamillaDSP) is an additional/intermediate sampling step. So in the case of say a 44.1kHz input file, either CamillaDSP is implementing a 22.05kHz filter or you DAC is, there is no way to disable it (unless you’re using a NOS DAC of course), you’re just changing where that filter is being implemented.
There is no audible jitter, even the DACs in cheap consumer devices from 25-30 years ago had “jitter/phasenoise” a hundred or more times below audibility.
But it’s not completely irrelevant, it might be irrelevant to consumer playback but not to engineers. We need to know the performance of say a DAC, even though artefacts are inaudible and therefore completely irrelevant to consumers, because certain workflows can require the mix (or parts of the mix) to pass through a DAC multiple times, thereby multiplying the artefacts.
How does it stop? Do you have a switch bolted to your skull so you can deactivate the parts of your brain that create “imagination”?
As you obviously cannot switch off your “imagination” and you have not done controlled testing (which eliminates cognitive biases) then you cannot know and again, “if you don’t know, then isn’t asserting you heard a difference dishonest?”!
Why? If everything “stays for weeks exactly the same”, including your knowledge/experience, belief and cognitive bias (expectation and others), then why would your brain ever calculate a different result, let alone “change day over day”? Some perceptual errors stay with us our entire lives (the stereo illusion for example) others are affected by cognitive bias and those can of course change with knowledge, experience and beliefs, sometimes over the course of years and sometimes just minutes.
Then how are you writing your posts? If your brain does not “save the setting” of say the rules of language (grammar and spelling etc.) then how could you ever communicate with anyone? Surely you’re not claiming you have no memory? When you listen to one of your favourite recordings, do you honestly have no recollection of ever having heard it before?
Yes I did, in fact I devoted a lengthy paragraph to it and you quoted some of it!
Why?well if there is a audible difference AND it can be measured then we should question the audiblity
With a minimum phase filter either you start the transition band at a much lower frequency (and have a much wider transition band) or you will have serious phase issues. What’s unusual in this case is that minimum phase filters are rare (in this application) but when you do find them, they rarely (if ever) sacrifice phase at such a low frequency.
Minimum phase filters should never be used in a DAC, except potentially for higher sampling rates where we can have a wide transition band that starts beyond the threshold of audibility. I’ve personally never seen a minimum phase filter implemented in a DAC that sacrifices phase at such a low freq for such a narrow transition band and there are almost no DACs that implement a minimum phase filter in the first place (except as a non-default switchable option). You don’t even find a minimum phase filter option in most DAW software, they just have a typical, near ideal linear phase filter built-in. Even in specialist resampling software it’s very rare to find a minimum phase option with phase issues starting at such a low frequency.
It’s interesting from the point of view that here we have a pair of anti-image filters, one with a relatively massive phase artefact (well within the audible spectrum), and the other with a relatively massive amount/duration of pre-ringing, yet a high proportion of test subjects thought the difference was minimal or non-existent and virtually none of the analysis demonstrated statistical significance! Besides this, the study doesn’t really tell us anything, unless you’re specifically using the SoX software with these highly unusual filter designs.
Firstly, the quoted archimago study was a preference study, not a threshold study. But even with preference studies, it is sometimes possible to get fairly conclusive results. However, you need large sample sizes and although quite conclusive, they don’t provide a precise result, they provide an average or a range.
well i would really like to solve the mystery but i can only guess here what seems most logical, the anti-image filter of the dac is worse then the ones implemented by software resampling and thats why i prefer resampling in camilladsp/pipwire with highest quality settings, even if it adds another "reconstruction step"
well i just tried recently with the ian canada hat, i bought a accusilicon clock instead of the stock NDK one, there was a audible difference/improvement
well, if we account for emotions etc atleast in theory it should be somewhat random how we perceive sound, and to some degree this is true but not to the degree where its "completely random" imo
1. What mystery are you trying to solve? If you’re talking about the mystery of why you’re perceiving something which shouldn’t be anywhere near audible, then there’s a test specifically designed for exactly this (a DBT). So if you “would really like to solve the mystery” then the obvious thing to do would be to use that test. If you don’t, then you obviously do not really want to solve the mystery!
There’s only two options: Either there was an audible improvement, in which case the DAC you’re fitting it to must have been broken/faulty, or there was not an audible difference and you’re just imagining a difference/improvement. The logical guess would be the latter, as a DAC broken/faulty in this way is extremely unlikely.
Resampling to 176.4 in your computer and then resampling again in your DAC does not make resampling easier, it makes it more complex. And, whatever you do, you only need one highly accurate clock (clock signal) in the DAC and as far as I’m aware, all DACs do.
Exactly my point, “if we account for emotions” then obviously they can change, as obviously so will knowledge and experience, therefore everything is not the same and our cognitive biases and perception can change! What it changes into is obviously not “completely random” we don’t suddenly or gradually stop hearing music or sound and instead just hear noise.
https://msbtechnology.com/dacs/clock-options/
Yep, there’s more than one way to skin a cat. Different DAC chip manufacturers do it differently, some require a 24mHz clock and derive the oversampled rates from that. I seem to remember that Analog Device chips used a variable rate clock (or did at one time) others allow two different clock input rates. As far as I’m aware you do not NEED more than 1 clock.
I think I have seen you do this more than once (not just an innocent typo): "mHz" means "millihertz." Hence 24 mHz = 0.024 Hz. Megahertz on the other hand is written MHz. Sure, the context tells us here that it can't be milli-, but it is good to be careful, especially when you are one of the authority figures here people know to take seriously (often for a good reason).
It is indeed “dumb”! Unfortunately, most audiophiles clearly don’t realise that it’s them who are on the “dumb” side of the argument. They are not entirely to blame, as there are significant financial incentives to keep them on the dumb side of the argument, although a little critical thought would often be enough to see through this false marketing. There is more than one avenue to address the quoted claims:
Or rather only two speakers typically at ±30° angle to the listener or perhaps headphones... The terms "hard left" and "hard right" are for the stereophonic signal and in that world there's of course a lot more than only "hard left" and/or "hard right." For example 75 % left and 22 % right. Just trying to make distinction between the properties of stereophonic signal and the physical properties of the sound reproduction systems.
