Need for a general statement?

[Theo]

There is a significant number of posts in this forum which lead to the statement that if the products don't measure differently then they should sound the same. It's becoming kind of redundant...
Is there a way to state this as a strong postulate which would avoid repeating this indefinitely? Like:
"Measuring no difference greater than 100dB within the 20-20kHz FR in the signal reaching your ears means that there is no audible change to hear".
Feel free to comment on the threshold or FR values (I chose a round and conservative one) or a different wording, there should be a something we could all agree on, shouldn't we?
This aims to be a minimum value that would not mean that a greater difference is necessarily audible to everybody.
The measure could be a simple substraction... or a Noise + Distortion value.

 

[Grave]

I think it would be hard to agree on, but:

• Frequency Response 20hz – 19 Khz within +/- 0.1 dB (Most DACs, due to the Nyquist limit of 22 Khz, start to roll off past 19 Khz when operating at 44 Khz sampling rate—the ODAC is down about 0.4 dB at 20 Khz). The widely accepted, but less conservative standard is +/- 0.5 dB (1 dB total variation) from 20 hz to 20 Khz.
• All Harmonic, IMD, Alias, Modulation, & Crosstalk Components Below –90 dBFS and total sum below –80 dBFS (0.01%)
• All Noise Components below –110 dB and total sum below –100 dBFS
• All Jitter Components below –110 dB and total sum below -100 dBFS

-NwAvGuy

I have been told these values are generous and much worse values should still be inaudible. I am not sure I believe that but w/e.

 

[Blumlein 88]

You'll probably have trouble getting agreement for sure. I once proposed at least ranking DACs into categories of what is essentially spurious free dynamic range. You'd have to test for flat response (+/- .1 db is needed here I believe). You'd have to look at total noise. Then you could post something like the spectrogram of sweeps I've used. That is a graphical representation of SFDR. You could use an FFT and pick whichever spurious signal is the highest. Put you'd need that for several frequencies. The only confounding factor there is close in sidebands on jitter, how high do you let those go and ignore, and at what distance from the test frequency.

Here is an example sweep with the background going to dark gray at -100 db. Single tone sweep left and dual tone IMD sweep right. The left hand pane has nothing above -100 db while the right hand begins to show some 3rd order distortion just barely above that level for the upper frequencies.

Here is a different bit of gear with same settings. There is some imaging there I didn't label. Some of these spurious signals are just above -70 db.

You can set the level at which things are ignored wherever you like. Here is the first device with the background going dark gray at - 120 dbFS. This device is still pretty clean even at this level. Ignore the horizontal line as it is an ADC artifact.

 

[Theo]

I was thinking in the line of a relative measure rather than an absolute one, which is more difficult to agree on as you rightly point out.
That is to answer the question of the existence of a difference rather than finding an absolute quality.

 

[Sal1950]

here is a significant number of posts in this forum which lead to the statement that if the products don't measure differently then they should sound the same. It's becoming kind of redundant...

It's not that I'm against the idea but,
Do you think that any statement will stop the believers from posting?,

But I can hear it
But you can't measure everything I can hear
All blind listening tests are flawed in some way and not really revealing of what I can or can't hear.

Believers are Believers and will fight to the death supporting the audible differences between a $50 and $5000 AC cord or USB cable no matter how idiotic the premise.

I saw David Copperfield make the 747 disappear with my own eyes.

 

[sergeauckland]

I think it would be hard to agree on, but:

• Frequency Response 20hz – 19 Khz within +/- 0.1 dB (Most DACs, due to the Nyquist limit of 22 Khz, start to roll off past 19 Khz when operating at 44 Khz sampling rate—the ODAC is down about 0.4 dB at 20 Khz). The widely accepted, but less conservative standard is +/- 0.5 dB (1 dB total variation) from 20 hz to 20 Khz.
• All Harmonic, IMD, Alias, Modulation, & Crosstalk Components Below –90 dBFS and total sum below –80 dBFS (0.01%)
• All Noise Components below –110 dB and total sum below –100 dBFS
• All Jitter Components below –110 dB and total sum below -100 dBFS

-NwAvGuy

I have been told these values are generous and much worse values should still be inaudible. I am not sure I believe that but w/e.

When I trained, values MUCH worse were considered inaudible, like distortion at -60dB 0.1% at any frequency and level under clipping and any permitted load.

Noise of 60dB below programme is generally inaudible, but becomes audible once programme is muted, or indeed during quiet bits, but certainly -80 Is more than good enough. Frequency response rather depends on the curve shape, but if I had something within 1dB 20Hz - 20kHz, I wouldn't be bothered at all. A 1dB wide (low Q) boost at, say, 2kHz may just be audible compared with a perfectly flat response, but the normal droop at LF and HF is pretty much inaudible.

That modern kit can do so much better is one reason why people have largely given up with specs, and it's all about touchy-feely emotional response rather than the numbers.

S

 

[Grave]

I was thinking in the line of a relative measure rather than an absolute one, which is more difficult to agree on as you rightly point out.
That is to answer the question of the existence of a difference rather than finding an absolute quality.

"All well designed DAC/amps sound the same."

 

[sergeauckland]

IDK, personally I would want everything to be a 96 dB or higher/lower depending on how you are looking at it.


"All well designed DAC/amps sound the same."

Yes, and generally with modern kit, it is. It's hard for it not to be that good, unless heroically badly designed or deliberate. If I had a modern amplifier with as much as 0.1% THD, I would look for a fault, even though that 0.1% is in itself inaudible, but could be a symptom of one or more problems.

I think we have all got so used to electronics being way over specced for transparency that anything less than 'excellent' is considered poor, even if inaudible.

When I work on vintage equipment, especially tape and doubly especially valves, I have to recalibrate my sense of what's good as distortions and noise levels are so much worse than with modern kit.

S

 

[Theo]

you think that any statement will stop the believers from posting?

No I don't think so, unfortunately pseudo-science will stay the plague which we observe everywhere. However, I was hoping that we could, at least among those who know what a measurement means, agree on some basic rule which we could use as a reference when trying to contain the flow of audiofoolishness... I know, I'm dreaming but isn't it worth the try?

 

[Blumlein 88]

Many distortions are apparently inaudible in general use if they are -60 db below the signal. I would think if all effects were less than - 80 db you are safe to say the gear is transparent. But that is end to end, the final output to the speaker. Most of which is a moot point, as other than speakers rarely will gear fail to meet that. Yet there are exceptions, it is still worth keeping makers honest.

Most of the time, if two units measure well and sound different level matched it will be frequency response. Typically at one of the extremes of course. Then those real though minor differences get blown out of proportion.

 

[Sal1950]

No I don't think so, unfortunately pseudo-science will stay the plague which we observe everywhere. However, I was hoping that we could, at least among those who know what a measurement means, agree on some basic rule which we could use as a reference when trying to contain the flow of audiofoolishness... I know, I'm dreaming but isn't it worth the try?

TBH, I'm more for a heavier finger on the post delete or member ban button by the mods.
Honest debate or folks really interested in learning is great, but we have the illusionist/delusionist cult that come here and try to beat that same ole drum we've heard for decades now. They'll post page after page of the usual drivel and circular arguements without a single new thing being said or presented that hasn't been posted here a hundred times before, millions at places like WBF or HeadFi to name two.
I've been the victim of heavy censorship from the other side on the believer sites and don't like it any better than anyone else. But in a large majority of cases when one of the cultist shows up here vomiting up the same ole, same ole, we can see where it's heading within a couple posts. Maybe a combination of a strong mission statement/rules and then inforcement by moderation would be a good approach.
Ever read the stuff like this at HeadFi
"Due to the flame wars that erupt as a result, this, and the other forums (other than the Sound Science forum), are DBT and ABX-free zones and posts about either will be moved or deleted."
Something like that here from our viewpoint might be appropriate.

 

[Grave]

Hopefully this forum will not not be like hydrogenaudio or head-fi. I was banned from both forums for insane reasons. I hate all censorship and I hate the cowards who hide behind it.

 

[Sal1950]

Hopefully this forum will not not be like hydrogenaudio or head-fi. I was banned from both forums for insane reasons. I hate all censorship and I hate the cowards who hide behind it.

I agree, but think we could find a middle ground somewhere in there.
No sense in letting a bunch of believer nonsense take over thread after thread as can so often happen.

 

[andreasmaaan]

There is an interesting and well-conducted study floating round in which under double blind conditions participants were able to reliably detect harmonic distortion through headphones to -94dB if the harmonic profile was obnoxious enough. This figure is surprisingly low compared to other studies I’ve seen, but the reason for this comes down to the design of the experiments if I recall correctly. Will dig it up and post it if I can later today...

 

[Blumlein 88]

There is an interesting and well-conducted study floating round in which under double blind conditions participants were able to reliably detect harmonic distortion through headphones to -94dB if the harmonic profile was obnoxious enough. This figure is surprisingly low compared to other studies I’ve seen, but the reason for this comes down to the design of the experiments if I recall correctly. Will dig it up and post it if I can later today...

I'd like to read about that.

So what like the 11th harmonic of 440hz or something like that?

 

[andreasmaaan]

Yes, and generally with modern kit, it is. It's hard for it not to be that good, unless heroically badly designed or deliberate. If I had a modern amplifier with as much as 0.1% THD, I would look for a fault, even though that 0.1% is in itself inaudible, but could be a symptom of one or more problems.

I think we have all got so used to electronics being way over specced for transparency that anything less than 'excellent' is considered poor, even if inaudible.

When I work on vintage equipment, especially tape and doubly especially valves, I have to recalibrate my sense of what's good as distortions and noise levels are so much worse than with modern kit.

S

I wouldn’t be confident that 0.1% would be inaudible. Probably yes if it’s all 2nd it 3rd harmonic, probably not if higher order.

 

[andreasmaaan]

I'd like to read about that.

So what like the 11th harmonic of 440hz or something like that?

It didn’t go into specifics unfortunately. If I recall correctly, the took the distortion profiles of opamps and caps and modelled these digitally. I don’t think they commented on what specific harmonics each model contained, but based on other studies (I’ll post another really interesting one I know of later today comparing alternative metrics to THD), I assume that the lowest levels of THD were detectable where the profile was predominantly upper order.

 

[Wombat]

I agree, but think we could find a middle ground somewhere in there.
No sense in letting a bunch of believer nonsense take over thread after thread as can so often happen.

Some basic science-based truths re established electrical measurement and otic and perceptual knowledge could be put up front as forum basics.

Those who participate are welcome to put forward credible(supported) agreement, enhancement or disagreement. If the support is not provided the discussion can be moderated accordingly, even moving posts to the Testing Area to allow free-speech.

There are too many wishful thinking/speculative threads/posts that lengthily go nowhere to the detriment of the forum's aims, IMNSHO.

 

[andreasmaaan]

Perhaps so as not to dilute this thread, I'll start a new thread later today looking at these two distortion studies I mentioned..

 

[Wombat]

I wouldn’t be confident that 0.1% would be inaudible. Probably yes if it’s all 2nd it 3rd harmonic, probably not if higher order.

Do your homework and increase your confidence(supportable) one way or the other. Opinion is not worth much in real terms re hearing in the broader arena.