SY--do you have access to a *good* ADC such that you might be able to do something like a multi-generational loopback? It would be an interesting test for us back home to have a few single-blind files that anyone who ISN'T using a DAW to break the files down and backing out the filter effect would be able to play along at home (I'm sure someone will tell me that my DAC at the very end will ruin this test, but that's why you do multi-generational to amplify the filters effect).
Food for thought.
Food for thought.
I'm by no means an expert in measuring DACs. Rather I was wanting to play around with ARTA's trigger/capture capability to see what I might learn about the Legato Link filter. Besides the Pioneer player, I also had an Onkyo with 5 selectable filters and a Muse Audio NOS DAC.
I made some *.wav files of a 1-bit impulse, tone bursts, and 18.5kHz/19.5kHz two tone test to write to a CD.
Attached are some highlights.
Obviously the frequency response and the impulse response are directly related to each other...difference sides of the same coin. Nevertheless, it is interesting to see them both. The IMD for the two tone test was an eye-opener. Check out that Muse-Audio! The IMD was easy to hear (especially the 1kHz) but the two test tones were not audible even though they were 50dB louder.
I'm sure the tests SY has in mind are much more detailed and carefully controlled. But, if anybody would like the *.wav files or details how to capture the measurements with ARTA let me know. Perhaps these type of simple tests could shed some light on what that Post-DAC filter is or isn't doing to the signal linear/non-linear response.
Last edited:
Good to see you here Steve.
Yes, the above blue quote was from Steve, didn't want to name him without his OK.
On the listening tests by SY, maybe be can work through that and I am for that, make no mistake. But I think Steve is giving us something objective to look at and perhaps pointing to the mechanism that explains 'anecdotally' what we are hearing fairly... ahem... emphatically and that we are not just imagining it. Finding that mechanism and potential cause also now becomes added motivation; that means we can collectively come up with a listening tests that Stuart, SY, wants.
I think also, before one conducts such a listening test, human bias can even intrude because we want to, or expect to, not to hear a difference. But when we have actual data that seems to indicate something anomalous and potentially audible, then the motive for those listening tests becomes a positive - and now we can a greater confidence, or let me put it this way, albeit bluntly: I am putting my hopes for a good outcome into the hands of a relentless pursuer, that is YOU, Stuart.
So I think that should be kept in mind, it is my head that is on the chopping block, so be careful with any sharp instruments, OK?
But, before we arrange for those listening tests, let's see if we can get data with the suggested post-DAC filter of this thread is 'on' and 'off' - that would be a good step first.
As I keep pointing out, I am in neither the purist objectivist camp nor the looney subjectivist camp. I would like to think that applies to most of us.
Cheers, Joe
PS: Look at the last two pages of Steve's PDF, the Muse DAC... that just has to be audible, hope so. Steve is right to point at those two.
Last edited:
Thanks. I have consulted the CS4362 pcb and the track and size are even smaller than I could remember. Impossible for my eyes and tools.
Tried to find other candidates. Haven't checked/searched the pdf but I found two SM5872 and PCM69AP. The former might not be suitable as I think they are "pure" DAC. The latter is I think too lowly, one bit or something, but it looks simple and easy to work with.
Another candidate is PCM1710u. Three chips on board. The chip is big but the trace and other components are as small as CS4362 pcb, so no go.
Attachments
SM5872 is suitable!
The SM5872 is very suitable for this approach. No tracks should be cut. As if the circuit has already implemented the same idea as Joe's.
Of course, unfortunately I cannot measure the value of the cap.
EDIT: Oh it was not the same.
The SM5872 is very suitable for this approach. No tracks should be cut. As if the circuit has already implemented the same idea as Joe's.
Of course, unfortunately I cannot measure the value of the cap.
EDIT: Oh it was not the same.
Attachments
Last edited:
Like I said, you can choose the listener. I'm sure you can find a volunteer.
I will not promise that the outcome is good, bad, or indifferent. That's not how experiments work. I can promise that the outcome is the outcome and the data and test will be fully disclosed and put out for anyone to try to replicate. That's how I've always operated.
Time to put your hypothesis to the test.
No, not the same thing. A hypothesis is a speculative guess that has yet to be tested. Example: I change X in a circuit, then believe that the sound is different. My hypothesis is that the change in X changed the sound. But it's just a hypothesis. However, like a good hypothesis, it makes a testable prediction which can be experimentally falsified. That's how we define "good" in the hypothetical sense: not that it's true, but that it's testable and falsifiable.
An example of a poor hypothesis is "The earth goes around the sun because Zeus wills it to do so." There's no experiment that can be designed to show that Zeus has nothing to do with it. Other examples of bad hypotheses include Sagan's famous "Dragon in the garage."
An example of a poor hypothesis is "The earth goes around the sun because Zeus wills it to do so." There's no experiment that can be designed to show that Zeus has nothing to do with it. Other examples of bad hypotheses include Sagan's famous "Dragon in the garage."
It is well known fact that different people have different understandings wrt to such attributes (especially when not listening in the same room to the same reproduction) but to conclude that it equally "undefined" as "nice" is not true.
For example, listeners will most likely agree that "soundstage" is related to spatial impression but not to the phrase "sound balance" (although a different sound balance can influence the perception of "soundstage")
But if someone wants to get more conformity he has train listeners by providing music samples wich illustrate the variation in the main parameters used for evaluation.
The EBU published some recommendations:
https://tech.ebu.ch/docs/tech/tech3286.pdf
Perhaps it could help to provide a link to another test with fully disclosed data and test details?
I seem to recall that you have never disclosed anything about the setup of any test you've ever done nor commercial affiliations, yet you have consistently claimed to have found all sorts of marvelous things. Without actually presenting your data.
Sorry, I work honestly and in the open.
I would look into the source of the 1K IMD more closely, an anti-imaging filter would not remove it so I would presume it is plain old analog IMD somewhere.
From reading his replies, i got the impression that Joe doesn´t know what in a real sensory test (preperation, setup, test protocol and statistical analysis) is done, so i´d thought, it could help if you provide a link to a detailed description. Basically it wouldn´t matter if it is one of yours or another one of methodologically sound quality.
Wrt to my tests, your recollection does not serve you well. I´m sure that we both hadn´t really published something done in our youth (i.e. roughly 35 years ago) and of course i´ve provided measurements and description of test procedure for one of the more recent (i.e. a preamplifier test) at diyaudio.
But now, as you´ve mentioned it, i can´t recall to have seen a detailed description of your tests and i´d be interested to know which way you have handled the matter of positive controls (afair you refused to accept the necessity of this basic element)
Positive controls are used where appropriate and described when used. I appreciate your offer of help and will not hesitate to ask questions as they come up.
Please ask, you don't have to assume.
Read all about them way back, digested them, and ultimately found it is flawed in many ways that proponents rarely admit. So it in itself is a biased process and its outcomes usually conforms to present orthodoxy and hence is an impediment to progress, in fact it is seriously danger of being nothing more that a brake.
I consider it of limited value, not valueless. It does not always translate.
I actually use another method, much simpler, but one that does not have audio 'establishment' imprimatur - and not that I care about that, it brings results that rarely requires statistical analysis (right there, a red flag should be raised) and is unambiguous. What the method is should not be hard to guess.
Cheers, Joe
Controls so that the tests are valid are not negotiable. No peeking, ears only. You pick the listener, it's up to you guys to pick the source material. As Steve said, "I'm sure the tests SY has in mind are much more detailed and carefully controlled."
Time to step up and put your hypothesis to the test. I've met you far more than halfway.
- Status
- Not open for further replies.