Is SINAD important? - "Myths" about measurements! [Video YT]

[phoenixdogfan]

Is the point of this that measuring SINAD and using it as a filter to determine which gear is worthwhile is an invalid way to approach selecting audio gear, and we should just rely on the golden ears of Cameron and Resolve and DMS instead? Really?

Resolve did this same shite about two years ago, and it was based as near as I could tell on the proposition that all forms of noise and distortion are not equal, so a set with identical SINAD but a higher percentage of noise than distortion would theoretically sound worse than the other way around. My point then and now is once you get past 100 db of SINAD, it's not going to make much if any audible difference at all. And since almost all the gear being measured today including amps and dacs now clears that bar, there's no reason to expect you'll need to listen to identically measuring with respect to SINAD units to detect something audible. In other words, once a certain level is achieved, it's going to be transparent irrespective of the mix between noise and distortions.

And yes, I do think it's little more than an attempt to generate clicks and revenue by starting a fight and trying to "piss out their territory."

 

[amirm]

That SINAD figure alone does not describe the harmonic structure nor how the harmonic structure will line up with our individual preferences.

The SINAD in my measurements are part of a full FFT of the distortion and noise of the product in the dashboard:

You can then easily compare the level of spikes, in this example -130 dB and lower, and SINAD to realize that it is noise, not distortion bound. I routinely make a comment like this in the review:

"Distortion is vanishingly small at -130 dB, with SINAD then being noise dominated. "

If X + Y = Z and I tell you that Y is much smaller than X, then you can approximate X = Y. Sometimes the two numbers are close in which case, you then look at the dynamic range measurements:

You see that the DNR is very close to SINAD. It is 2 dB better because those distortion products do add up to something.

If you don't know these things, then please watch the full tutorial I have given on this, which also follows my AES paper and talk on the same:

No DAC is rated precisely on SINAD either as I always put them in one of four buckets:

Now, if someone wants the best of the best SINAD value and the gear is still cheap enough for them to afford it, then that is cool for them to purchase. I personally use a DAC that is in the middle of the blue range.

I have not watched his video but if this is all about, all the information and knowledge is here.

 

[raif71]

GoldenSound talks through some myths (according to them) and misconceptions about audio/hifi measurements.

The Headphone Show

NB FOLLOWING SUMMARY ADDED BY MODS COPIED FROM A POST DOWN THREAD FROM THE OP .

The video is disappointing in reality,

but I thought it could open up debate In fact, the conclusion is "don't rely only on measurements" and "don't rely only on the ears"

Thanks for pointing out this video. Gonna watch it on youtube later...

 

[amirm]

Man did those three clips sound horrible. The notion that the first one represents a tube amp (2nd harmonic) and folks prefer it so it must be good, is just false. It wouldn't take much or any expense to put people through listening tests of that clip vs the original and see if people prefer the distorted 57 dB clip.

I also really dislike comparing three distortions without also presenting the original. Maybe the original is so distorted that the impairments don't matter. Or that the original is much better and folks listening blind could say all three suck, in which case the claim about tube sound would be soundly shown to be false. Pun intended.

 

[amirm]

The notion that SINAD is not a predictor of subjective experience is plain false. Yes, analyzing the distortion profile and arriving at audibility levels is hard (not impossible given the spectrum as I show in FFT). However, the noise aspect is trivial to assess. Same with distortion products at the extreme. If these are below threshold of audibility, then it doesn't matter what they are. They device is transparent to the input. I do this analysis routinely and is based 100% on psychoacoustics and listening research.

So I test one device and it falls in the above category of transparency. Then I test another device that costs 5X more and is above that level. Isn't this all that you need to decided on a purchase between these two?

 

[Holmz]

The notion that SINAD is not a predictor of subjective experience is plain false. Yes, analyzing the distortion profile and arriving at audibility levels is hard (not impossible given the spectrum as I show in FFT). However, the noise aspect is trivial to assess. Same with distortion products at the extreme. If these are below threshold of audibility, then it doesn't matter what they are. They device is transparent to the input. I do this analysis routinely and is based 100% on psychoacoustics and listening research.

So I test one device and it falls in the above category of transparency. Then I test another device that costs 5X more and is above that level. Isn't this all that you need to decided on a purchase between these two?

In his video he pretty much said that looking at 2 devices separated by 1 or 2 dB of SINAD was meaningless using solely SINAD.

…
If X + Y = Z and I tell you that Y is much smaller than X, then you can approximate X = Y. Sometimes the two numbers are close in which case, you then look at the dynamic range measurements:

I think you mean X ~= Z ?

I have not watched his video but if this is all about, all the information and knowledge is here.

It may be somewhat more ubiquitous than that.

 

[MyCuriosity]

I found the video very informative and helpful.

 

[MyCuriosity]

The notion that SINAD is not a predictor of subjective experience is plain false.

I don’t think he implied that. What I understood Is that SINAD is a useful measurement but will not always tell you if one item sounds better or worst than other

 

[JSmith]

One major goal of signal amplification is low noise and distortion.

Thus there is no reason to not get a device with low noise and distortion... since the goal here is high fidelity.

However there are many aspects to device purchase to consider.

No one has ever said buy on one SINAD figure alone... that is very different to using a measurement metric to create a ranking list.

Cameron has many sponsors.... so an "informative" video is still advertising/views.


JSmith

 

[solderdude]

Man did those three clips sound horrible. The notion that the first one represents a tube amp (2nd harmonic) and folks prefer it so it must be good, is just false. It wouldn't take much or any expense to put people through listening tests of that clip vs the original and see if people prefer the distorted 57 dB clip.

I also really dislike comparing three distortions without also presenting the original. Maybe the original is so distorted that the impairments don't matter. Or that the original is much better and folks listening blind could say all three suck, in which case the claim about tube sound would be soundly shown to be false. Pun intended.

In the 3 posted cases it is obvious SINAD matters (when it is really low, say bottom of Amir's SINAD chart) but is 'masked' by the absence of a >90 SINAD reference.
It would have been great (to make his point) if a higher SINAD sample was included to make a point about SINAD having some relevance.
And even there he could have done a 100 SINAD comparison between a rolled-off and not rolled-off sample to make his point.
Missed opportunity. This could have easily fitted in the video at the spots where he came back to the same mantra over and over again...

Please do the video again Cameron (I suspect he is reading this) but this time explain why SINAD also can show its 'value' (but can be rather pointless above certain numbers in certain cases when signal fidelity is of importance).
A very high SINAD, for instance, can be of importance when digital volume control is used (with power amps fully open) and/or when very sensitive (horn) speakers or head-/ear-phones are used.
Surely you can warn people not to fuss over the last digit in the SINAD.

 

[Waxx]

There are many of this kind of video's and discussions going on on the internet, on all kind of fora and platforms. It's a sign that ASR has an impact and that people start to discuss real specs more than all the woo woo that was there before. At least now they talk about measurements, before that was ignored or unknown. And a lot of people (reviewers, hifi freaks and so) feel treatened by this, and need to attack it. But a lot also support the ASR view and it did and still does change things in the hifi world in a good way. Or they make a clickbait video like this in the OP post as this is a heaten subject now that draw views....

And i personally SINAD is only one factor that tells what a device does, that is why Amir measures way more than noise alone. And once a sinad is above 96dB (16Bit audio) it's not relevant anymore for sound, only relevant for technical comparisons in reality. But other factors are equal important for me than that, and if you got a Sinad below 96dB it's important to look at the distortion profile, because a lot of good sounding amps are not that noisefree, but the noise does not spoil the sound (tube amps and so) for many. Reducing a technical review to a SINAD number is very ignorant i think, it's the total package with all measured specs that tell you really what a device does, and SiNAD is only one factor.

 

[antcollinet]

I'm aware of that. But I'd like members who find and post "interesting" YouTube videos, especially of this nature, to tell us why it may be of interest /controversial / worthy of debate so we can choose to watch or not . Not have to watch to decide if it's debate worthy .

The click baity nature of the captions on the thumbnail was the clue . The gasket blowing comment was confirmation of how some folk want this to play out.

This. So much this. In fact can we add it as guidance in the Terms page?

I'd also like them to post their views on the content. That is how to start a debate - not ask people to start debating with some rando on youtube who is not even here to participate.

 

[antcollinet]

But it is one number on the ranking chart.

Well if you're going to have a ranking you need a number. Which one would you pick?


As far as I am concerned the sinad based ranking chart is fine as input into a shortlist of devices for consideration - or as a way of rejecting 1/2 the devices you don't need to look at.

 

[antcollinet]

So, lots of this discussion is beyond my level of understanding. But as near I can tell, this guy claims two rigs with SINAD of about 50 can sound different. Is that right?

Since my rig has a SINAD of 90 or better, I'm guessing it's not relevant for my listening. Is that right?

TIA

Correct.

 

[Holmz]

Well if you're going to have a ranking you need a number. Which one would you pick?

Probably the one with the features I want.

As far as I am concerned the sinad based ranking chart is fine as input into a shortlist of devices for consideration - or as a way of rejecting 1/2 the devices you don't need to look at.

Yes - correct.

And I am not going to sell a unit at #10 or #20 to replace with a unit at #1.
(Especially if that unit at #1 is a Topping device.)

 

[antcollinet]

I'm not sure I follow. What's the point of engineers achieving a low THD in their product if it's of no benefit to the end consumer?

Because low THD is a benefit. Once it is low enough (as in any current halfway decent electronics) it becomes inaudible.

THD as a published measurement is only useful to say if it is audible, or inaudible. Once it becomes audible, it tells you nothing about what that audible distortion will sound like. That is where it lacks benefit - as a published metric, not as a design target.

 

[antcollinet]

(Especially if that unit at #1 is a Topping device.)

I hear you. Topping has a lot of work to do on convincing me on reliability, and more importantly, competent/local after sales support before I will go there.

 

[antcollinet]

What I understood Is that SINAD is a useful measurement but will not always tell you if one item sounds better or worst than other

Which is fine - but falls into the category of "no shit Sherlock"

In other words, no-one who understands the measurements done here has ever claimed that it will. His amazing message is a message already well understood and explained over and over again to people who come and ask.

That is why we have all the other measurements.

 

[Holmz]

Well if you're going to have a ranking you need a number. Which one would you pick?
…

I usually look at the harmonic structure first.
The golden eared crowd seems to gravitate towards amps which have 2nd harmonic louder than the 3rd, and the rest of the lice comb of harmonic very low.
There is probably something to that as the correlation is pretty strong, assuming that we ignore the SET and other tube and horn crowd.

After that it is noise level, as hissing grates on me.

Then the usual cost, features, and jewel/feng-shui stuff.

 

[olieb]

Well if you're going to have a ranking you need a number. Which one would you pick?

There are two big reasons to use SINAD: It is so super simple and it is easy to compute!
But otherwise I am not convinced to use SINAD as a ranking number for SQ. [I use it as some kind of safety number to be below the threshold of audibility.]

For ranking the performance I would like to include linear (FR) deviations, harmonic (and other non-linear) distortion and noise.
And these aspects would have to be weighted according to audibility.

Linear deviation (FR) becomes audible around 0.1dB or 1% of amplitude. But this depends on the frequency range and maybe weighting with ERB-scale can account for this. [And there is of course the width of the deviation.]

Harmonic distortion becomes audible around 0,1% (-60dB), but this depends on frequency and number (frequency) of the harmonic products. So we would need some kind of two-parameter weighting at least. Level dependency might have to be taken into account, too. I do not know of any robust science on that.
[Just weighing the harmonic products by Fletcher-Munson might be a first guess, but there is masking, which is more complicated.]
Other non-linear distortion (IMD) is important too, but it seems to be rather closely correlated to HD for almost all electronics (and speakers?)

Noise will become audible around 0,01% (-80dB), but again this depends on noise spectrum. Fletcher-Munson (A-weighting) seems to be the obvious choice.

All this are only ballpark numbers and depend on program and use case.
This kind of weighted sum would be a better choice in my point of view.

But now it is not so simple anymore, and there is no consensus, of course.

And the question is, how important is it anyways? In electronics it is not a big problem to stay below audibility (unless using tubes) and in speakers SINAD is not the number used at all.