What do you think makes NOS sound different?

That’s a big misconception.
The older I became the better and faster I could tell one system from another.
I can hear all kind of differences despite not having the ears of an 18 year old, so loss of HF is not as bad as you might think. Your brains are a very important part in what you hear.

Youngsters are generally not interested in this sort of things. To them sound quality is mostly no issue at all, volume is way more important.
So I see no reason for you not to participate.


I have participated a lot in these kind of tests. But my results were becoming more and more inconsistent. And therefore I think my own opinions on what I perceive don't matter for the rest of humanity.




When noise level is -80 dB, it doesn’t mean that hearing sound stops at this level.
Like an FFT, our hearing system has filter bins that are 3 to 4 Hz wide in the sensitive areas.
So when noise over a 10Khz BW has a level of -80dB, the level at the 3Khz bin is 10log(4Hz/10Khz) = circa 30 dB lower.
That’s why we can hear a 3khz tone at -100dB that is not masked in -80dB broadband noise.

Hans


I completely agree with this.


I just don't think its relevant when listening to normal music.


Masking is a real thing
AudioMaskingThreshold.png
And when listening to music, we don't just listen to one frequency, we listen to a whole spectrum of frequencies. And all these frequencies are masking a whole lot of other frequencies, lower in level. This is the reason why we can't hear artefacts below a certain level.
 
An other umpa lumpa of science 😀 (one of Sheldon Coopers best jokes)


I'm interested in good sound.
I look at what inhibits this for the most. I think this is speakers by a very large margin, so I'm focussing on that.

(...)

Modern commercial DAC's of a couple of hundred bucks perform at the theoretical limit of what is possible. Artefacts are 120dB below the full scale signal level. I just can't see how anyone can improve upon that, without serious cooling equipment.

OK, so your remark about amateurs making an improvement in post #1519 had nothing to do with DACs. As you posted it in a DAC thread, I assumed it had. Hence my confusion.
 
Last edited:
Bill, looking at the factual theme of this thread for someone convicted that a very low cost Dac does it all and cannot be improved sonically, could you explain why you are frequenting this thread.
It is obviously not a subject that keeps you busy and your contributions do not help in any way in finding the answers to Ken’s questions.

Hans
 
Hans is asking the right questions - I was also puzzled why Bill suddenly crashed this party with unhelpful posts. I sugget he goes back focusing on his speakers, as he rightly mentioned by the way.

we will stay on topic here - hope you will get some further results Hans on the listening test - before we forget what this thread was about.
 
At least he's right to say that the conclusion about the OS filter implementation being the main culprit stands on feet of clay because of the limited number of participants in the listening tests. And also because these test were not methodological enough (not speaking about the echo test in this case, Marcel's approch was as methodological as it can get).
 
I don't agree with this.


The test results in this thread can't make that claim.

Why not, Bill? I quite specifically included the verb, 'appears', to recognize that our findings have a very low statistical confidence. The findings are, statistically, highly questionable. Which is not the same as them being necessarily wrong. Who knows, a higher sample size may, or may not support the apparent findings. However, if you wish a different verb, how about, 'seems' or 'suggests'?

We all fully recognize that we are not running a scientifically controlled, AES approved research test. We are an ad hoc group of internationally dispersed hobbyists, of relatively limited practical means and experiment process control, attempting to explain a puzzling subjective phenomena which has long seemed to lack a clear explanation. As such, we are forced to endeavor within certain practical realities and limitations. Even so, when have such obstacles deterred the efforts, scientifically insufficient as they are, of the DIY audio community? Else, I suppose, we should all just shut up and accept whatever others tell us MUST sound perfect, even though our ear/brain informs us that it does not.

While we, indeed, are not able to scientifically prove our apparent group findings, that fact does not prevent individual contributors from concluding they have consequently identified a more subjectively satisfying 'system formula' for digital playback to their ears. Which, are the only ears that truly matter for a home HiFi system.
 
Last edited:
...And also because these test were not methodological enough (not speaking about the echo test in this case, Marcel's approch was as methodological as it can get).

Yes, Marcel's methodology and analysis are first rate. Something which we all should aspire to, yet most of us are probably not knowledgeable enough to do as well as he can. Just the same, his test was still saddled with any number of practical limitations which were out of his control, and which contribute to bring the test findings into question. Essentially, the same as they have for all our other experiments. Prime among these, is a lack of control over the listening system of each participant, in order ensure optimum experiment uniformity and validity. While I speculate that this is probably the greatest reason for the Echo experiment group results being inconclusive, it's certainly far from the only possible reason.
 
Last edited:
"lack of control over the listening system of each participant" is irrelevant IMO if the test (almost) produces a null result anyways. If something cannot be heard regardless of the system used, what's the point in ensuring system uniformity?
 
"lack of control over the listening system of each participant" is irrelevant IMO if the test (almost) produces a null result anyways. If something cannot be heard regardless of the system used, what's the point in ensuring system uniformity?
Whether the listening test is (bias & other variables) controlled or not is relevant. Participants are supposed to do a pre-test listening or at least they should be allowed to. It's the subjective listening sessions to see if they can discern a difference or not. If they do, then move on to controlled listening test.
 
"lack of control over the listening system of each participant" is irrelevant IMO if the test (almost) produces a null result anyways. If something cannot be heard regardless of the system used, what's the point in ensuring system uniformity?

Except, that's the very point. We don't know whether or not the lack of system control was THE reason for the null result. This is positively a concern for an experiment intended to test a possible causal mechanism for the subjective difference which many of us hear between common on-chip OS, and NOS.

The base case for test evaluation was assumed to be NOS DAC, which do not introduce the error being tested. The lack of control over every participant's listening system, however, resulted in all but one participant instead utilizing an OS DAC as the base case. Which introduced the error which we were explicitly hoping to isolate the perception of, in to all test file replay. All files played by an OS DAC introduced some base amount of the subject error.
 
Whether the listening test is (bias & other variables) controlled or not is relevant. Participants are supposed to do a pre-test listening or at least they should be allowed to. It's the subjective listening sessions to see if they can discern a difference or not. If they do, then move on to controlled listening test.

And if they don't hear a difference in the subjective listening test you exclude them from the controlled test? That doesn't sound right at all...
 
Last edited:
Except, that's the very point. We don't know whether or not the lack of system control was THE reason for the null result. This is positively a concern for an experiment intended to test a possible causal mechanism for the subjective difference which many of us hear between common on-chip OS, and NOS.

The base case for test evaluation was assumed to be NOS DAC, which do not introduce the error being tested. The lack of control over every participant's listening system, however, resulted in all but one participant instead utilizing an OS DAC as the base case. Which introduced the error which we were explicitly hoping to isolate the perception of, in to all test file replay. All files played by an OS DAC introduced some base amount of the subject error.

You have a point there.

Still makes you wonder why, when there are at least three guys driving this thread with NOS dacs (you, Hans and Doede), we wound up with only one of them taking the test... pretty poor attention 😉 Without the five guys with OS DACs you would have wound up with exactly one result. And that one with random/null outcome...

Enough of the bashing 😀
 
Tfive,
correction before accusing Ken and me of poor attention:
I don’t have a NOS but an OS Dac that automatically converts everything to 192K, including the 8* oversampling before the SDM.
So no matter whether I feed 44.1, 88.2 or 176.4, it will be all converter to 192K*8, quite inappropriate for this test because too many steps added to the process.
Ken’s Nos Dac is broken and what he has is similar to my Dac, exept that everything is converted to 96Khz.
But DDAC has indeed a Nos Dac.
This was all documented in previous postings.

Hans
 
Last edited:
OK, then I was not paying enough attention - sorry for that.

I thought that at least the OP would have a NOS DAC ready when suggesting these experiments instead of complaining of not enough participants with a NOS DAC after the test was conducted - but that's my personal logic 😉
 
You have a point there.

Still makes you wonder why, when there are at least three guys driving this thread with NOS dacs (you, Hans and Doede), we wound up with only one of them taking the test... pretty poor attention 😉 Without the five guys with OS DACs you would have wound up with exactly one result. And that one with random/null outcome...

Enough of the bashing 😀

Not taken as bashing, at all. You wonder the same thing which I do. Judging from the nearly 80K views, someone is still interested in the thread’s activities. I’ve already discussed the issue of limited participation offline with Hans, and it’s my suspicion that it’s largely due to simple experiment fatigue, but perhaps that suspicion is mistaken. I think that most of us tend to come and go on these threads. Dropping in and out, participating as our interests change, and limited time allows.

I think this principle probably goes double for the time and effort consuming exercise of conducting physical experiments. While your suspicions for why participation has become so limited may be different from mine, however, we all recognize that participation has slowed to essentially, crawling across the finish line. Regardless of the outcome of the 176.4 experiment, however, I already feel that the thread has successfully addressed the original investigation question. At least, at a level which allows an DIY DAC designer to functionally leverage the apparent indications. So, I think it fortunate that this is our final experiment, which a short time after finishing, I will conclude the overall investigation with a final report.
 
Have thought maybe the lack of participation is due how the thread is organised? at least some of data/info from the thread compiled in first post would make a big difference to accessibility. Anyone new to the thread (or who hasn't been following closely) has to sift to 100s of posts to get their bearings... it's a problem in many long threads.

I really fail to see how the original question was addressed at all, the original question is NOS vs OS, right? not 'high' vs 'low' quality OS? That is already a more technically discussed and less interesting question than the former I feel.

NOS dacs are required for the test to start with.
One thing that bothers me is that OS needs digital attenuation, NOS does not, but it is needed to make the test files. Do we assume digital attenuation is flawless and this is not a potential advantage for NOS?
 
Have thought maybe the lack of participation is due how the thread is organised? at least some of data/info from the thread compiled in first post would make a big difference to accessibility. Anyone new to the thread (or who hasn't been following closely) has to sift to 100s of posts to get their bearings... it's a problem in many long threads.

The thread is self-organized along a sequential time-line, like most other threads. You may have a point about it not being easy for newcomers to enter the thread, however, that is an issue confronting every other lengthy thread, isn’t it? Let’s be accurate, I have repeatedly posted brief summaries of our progress, and updated the organized suspect list outline, including adding my own comments for each item in italics, have I not? I also have placed an easy to notice relevant title on such posts. So, I really don’t know what more you are talking about. Perhaps, you have a suggestion?

I really fail to see how the original question was addressed at all, the original question is NOS vs OS, right? not 'high' vs 'low' quality OS? That is already a more technically discussed and less interesting question than the former I feel.

No, not really. Here is the first sentence, of the fourth paragraph from within my thread starting post #1; “ I want to make clear, I’m not expressing a set preference for NOS over OS. The issue for me is not that one sounds better than the other, but rather why they sound different at all?” That says it all. I’m unsure what you mistook to be the thread goal.