Could we all be wrong about SINAD?

[EB1000]

Hi

I’ve noticed that my NAD C658, which measure like cr@p, sounded much better than the Denon X4700H, which I’ve recently auditioned at my home, before and after running Audyssey.


Another example is the PS Audio Directstream DAC, which also scored terribly on ASR, and yet, most audiophiles say that it sounds awesome, despite the 80dB SINAD.


What if? What if NAD and PS Audio engineers deliberately added some odd and even harmonics, with specific magnitude ratios, so the precepted sound to the human ear will be optimized to have these audiophile characteristics, such as “warmth” and “depth”, which cannot be captured by the Audio precision analyzer? Maybe they did some research and came up with this secret recipe for an harmonic contamination pattern that does magic to the sound perception, but results in a cr@ppy SINAD?


What if we are all wrong here? Is it time to shut down ASR and lock Amir up?

Did anyone research this? Is there an AES article which correlates harmonic contaminations to sound perception?

This is worse looking into...

 

[AdamG]

I call dibs on the Panther collection adoption.

 

[Leporello]

No. The problem is the unspoken assumption that undistorted hifi sound reproduction must necessarily also be subjectively pleasing - if not there must then be something wrong with the theory and measurements. This of course is nonsense.

 

[virtua]

As badly as some devices measure here, most of them don't really have 2nd and 3rd harmonics shown in the FFT that would be easily audible. For example, the PS Audio DirectStream DAC only has a -80db spike on the third harmonic, which is much higher than the other spikes on the graph.

Some devices do go to recreate 2nd and 3rd harmonics for an effect like the Schiit Vali, Garage 1217 amplifiers, but even those might only peak at most -50db on 2nd harmonics at highest which is still quite low in level. But generally there is distortion that is understood as pleasing, and distortion that is understood as generally unpleasing. A lot of devices which seem to have bad SINAD definitely don't come across like it's being done in a tasteful way on purpose.

 

[voodooless]

As badly as some devices measure here, most of them don't really have 2nd and 3rd harmonics shown in the FFT that would be easily audible. For example, the PS Audio DirectStream DAC only has a -80db spike on the third harmonic, which is much higher than the other spikes on the graph.

Look at the multi tone measurement. It shows only 10 clean bits. I’m pretty sure that goes intonatie audible range of crappyness.

 

[virtua]

Look at the multi tone measurement. It shows only 10 clean bits. I’m pretty sure that goes intonatie audible range of crappyness.

Well I definitely didn't see that lol, still - I don't think that should be a characteristic of an amplifier with pleasing distortion characteristics. I might be wrong but there seems to be more going on there than just the 2nd and 3rd harmonic.

 

[abdo123]

Are you capable of level matching the gear and doing A/B testing and actually determine whether the sound difference between these 3 gears is actually true?

Not that i doubt you or anything, but these claims are (Gear X sounds better than Gear Y) almost a daily occurrence on this site. I have heard ridiculous testimonies of gear even I also own personally. It's just we need some sort of controls to make sure what you're describing is real. From that vantage point we can talk about it.

Even just level matching is usually enough to break people from their cognitive dissonance. Or push problems gear have forward like a dog's balls.

 

[MZKM]

Hi

I’ve noticed that my NAD C658, which measure like cr@p, sounded much better than the Denon X4700H, which I’ve recently auditioned at my home, before and after running Audyssey.

If you aren’t playing them back at the same SPL, then you can’t have solid finding. Next critique would be how quick you can switch between both amps.

Also, see Amir’s review of Audyssey, it strips your system of bass, you sadly need the $20 add and give it a target curve which has rising bass as well as disables the MIdrange Compensation (MRC).

 

[charleski]

I’ve noticed that my NAD C658, which measure like cr@p, sounded much better than the Denon X4700H, which I’ve recently auditioned at my home, before and after running Audyssey.


Another example is the PS Audio Directstream DAC, which also scored terribly on ASR, and yet, most audiophiles say that it sounds awesome, despite the 80dB SINAD.

Amirm noted significant operational problems with both units in his reviews, though the problems with the NAD seem more severe and unavoidable. A failure to spot the difference subjectively between a SINAD of 90dB and 94dB is going to surprise no-one who understands what these measurements mean. As @abdo123 said, 'sounds much better' really doesn't mean anything without controls.

As for 'most audiophiles say', it's really just a case of 'most audiophiles are able to echo each other' - it's Stereophile class A+ after all!

 

[sergeauckland]

SINAD, or at least, THD, better than -60dB is inaudible under any natural circumstances with speech or music. It's possible to create an artificial situation whereby the lower harmonics are well below -60dB, but some higher harmonics, say 5th, 7th etc are at -60dB and these become audible. However, there's no natural method for an amplifier to create higher level harmonics at -60dB without the lower level harmonics, like 2nd and 3rd, being a LOT higher, so as far as I'm concerned, if an amplifier has THD levels of -60dB at all frequencies and levels under clipping, it will be completely transparent in normal use on speech and music.

In practice, if the THD is around -60-80dB, the noise element of SINAD is typically some 20dB lower so has very little influence on the SINAD. That's why I think that treating SINAD figures of -80dB as bad, is nonsense, unless the SINAD is made up predominantly of noise.

Combining distortion and noise into one figure may be helpful in marketing terms, not that most people understand it, but unless the THD and Noise are shown separately, it's pretty meaningless. The only thing one can say for SINAD, is that if it's, say, -100dB, then it doesn't matter one iota whether it's distortion or noise limited, neither will be audible. I just wish we gave up on SINAD and quoted the two numbers separately.

S.

 

[MakeMineVinyl]

Personally I'm very skeptical that SINAD has much value in actual listening. I consider it just a data point among a lot of other data points. Consider the fact that the speakers which all of us listen to have distortion which would be utterly, absolutely, and unquestionably horrible in any other electronic component. Never mind that all of our rooms add 'sound' to the signal in proportions which would also measure horribly in any objective way.

Yet people don't bitch about this form of distortion. To me, that's a major disconnect to obsess over one distortion and ignore another which is orders of magnitude worse.

 

[abdo123]

SINAD, or at least, THD, better than -60dB is inaudible under any natural circumstances with speech or music. It's possible to create an artificial situation whereby the lower harmonics are well below -60dB, but some higher harmonics, say 5th, 7th etc are at -60dB and these become audible. However, there's no natural method for an amplifier to create higher level harmonics at -60dB without the lower level harmonics, like 2nd and 3rd, being a LOT higher, so as far as I'm concerned, if an amplifier has THD levels of -60dB at all frequencies and levels under clipping, it will be completely transparent in normal use on speech and music.

In practice, if the THD is around -60-80dB, the noise element of SINAD is typically some 20dB lower so has very little influence on the SINAD. That's why I think that treating SINAD figures of -80dB as bad, is nonsense, unless the SINAD is made up predominantly of noise.

Combining distortion and noise into one figure may be helpful in marketing terms, not that most people understand it, but unless the THD and Noise are shown separately, it's pretty meaningless. The only thing one can say for SINAD, is that if it's, say, -100dB, then it doesn't matter one iota whether it's distortion or noise limited, neither will be audible. I just wish we gave up on SINAD and quoted the two numbers separately.

S.

I agree with your conclusion but I don’t agree with your noise threshold. I think it should be a bit lower still.

 

[virtua]

In practice, if the THD is around -60-80dB, the noise element of SINAD is typically some 20dB lower so has very little influence on the SINAD. That's why I think that treating SINAD figures of -80dB as bad, is nonsense, unless the SINAD is made up predominantly of noise.

Could not agree more, some amplifiers might have pronounced 2nd and 3rd harmonic peaks but a relatively low noise floor - resulting in a terrible SINAD number. Others might have lower peaks but terrible power supply noise rejection and other noise floor issues, but because the harmonic peaks aren't high it might show a better SINAD than the former. In practice the latter device will likely be much worse, and more likely to have audible noise issues even without playback. 2nd and 3rd harmonic distortion can be much higher in level and you can generally get away with it without much perceived negative affect on sound quality vs other kinds of distortion or noise.

Generally noise has been a much more noticable issue than distortion ever has been with the amplifiers that I've owned and listened to.

 

[Raindog123]

Here is a thought, rather trivial and obvious one, I recently shared on Auralic forum:

“The truth about modern stereo (ie two-channel!) equipment can possibly be distilled into three tenets:

1. Whether we like it/agree with it or not, from the “true to the source” point of view, modern audio electronics far exceed human hearing (and human audio processing) limits. As such, a “decent modern” set of the source-amplification-interconnect is indistinguishable from another “decent” modern set (in terms of sound quality — leaving out aesthetic, ergonomic, brand loyalty, etc. considerations).
2. Transducers (aka speakers) [due to underlying technology state-of-the-art] and their interaction with the listening environment (room) [due to topology uniqueness and relative complexity of “multi path” soundwaves-walls interaction] is the only audio component that can be noticeably audibly varied thus improved. So — again, “from the reproduction quality “true to the original” point of view — speakers and speaker-to-room optimization — room treatment, compensation/equalization software — are the main (only?) meaningful upgrades at some point.
3. After the above is said… Again and again it appears that we, the listeners, do not actually want “true to the source” sound. Often we like our sound “with character”, with slight imperfections: colored by tad of jitter, slightly distorted by harmonics, with nonlinear frequency response (emphasizing low’s… or high’s) As well as psychological factors, including eg price-tag-dazzle and VIP showroom treatment. From the component-upgrade point, this translates into (and is governed by) “personal preference”: listening, socializing, doing research, spending time with… As this is what ultimately makes [many of] us happy. So, just try various options, all over, and see for yourself! ”

 

[Colonel7]

I’ve noticed that my NAD C658, which measure like cr@p, sounded much better than the Denon X4700H, which I’ve recently auditioned at my home, before and after running Audyssey.

The NAD is a preamp/streamer isn't it? What is the amp you're comparing with the 4700?

 

[abdo123]

Here is a thought, rather trivial and obvious one, I recently shared on Auralic forum:

“The truth about modern stereo (ie two-channel!) equipment can possibly be distilled into three tenets:

1. Whether we like it/agree with it or not, from the “true to the source” point of view, modern audio electronics far exceed human hearing (and human audio processing) limits. As such, a “decent modern” set of the source-amplification-interconnect is indistinguishable from another “decent” modern set (in terms of sound quality — leaving out aesthetic, ergonomic, brand loyalty, etc. considerations).
2. Transducers (aka speakers) [due to underlying technology state-of-the-art] and their interaction with the listening environment (room) [due to topology uniqueness and relative complexity of “multi path” soundwaves-walls interaction] is the only audio component that can be noticeably audibly varied thus improved. So — again, “from the reproduction quality “true to the original” point of view — speakers and speaker-to-room optimization — room treatment, compensation/equalization software — are the main (only?) meaningful upgrades at some point.
3. After the above is said… Again and again it appears that we, the listeners, do not actually want “true to the source” sound. Often we like our sound “with character”, with slight imperfections: colored by tad of jitter, slightly distorted by harmonics, with nonlinear frequency response (emphasizing low’s… or high’s) As well as psychological factors, including eg price-tag-dazzle and VIP showroom treatment. From the component-upgrade point, this translates into (and is governed by) “personal preference”: listening, socializing, doing research, spending time with… As this is what ultimately makes [many of] us happy. So, just try various options, all over, and see for yourself! ”

and then we have people like @Pdxwayne detecting subsonic noise at -70 dBFS.

 

[DVDdoug]

despite the 80dB SINAD.

In many cases that's below audibility. In the analog days that would have been awesome!

will be optimized to have these audiophile characteristics, such as “warmth” and “depth”, which cannot be captured

You can't measure something that's not even defined. Apparently, "warmth" is distortion, which is different from what I used to think. And "audiophiles" sometimes "hear things" that they can't actually hear (in a blind listening test).

I try to stick to words with scientific/engineering definitions like, noise, distortion, and frequency response. Sometimes I'll say "dull" or "bright" (which are related to measurable frequency response or frequency content) but I try to avoid those too.

 

[amirm]

SINAD is low commonly due to following reasons:

1. The "designer" doesn't know what he is doing and has produced utter garbage.

2. The designer is competent but never measured so is not aware of obvious problems with the implementation. Had he measured, he could have fixed them at usually little to no cost. Because instrumentation is so expensive, this is extremely common.

3. The designer is resorting to "audiophile" concepts that have never been proven to actually benefit the sound. Examples include no feedback amplifiers, sticking tubes in otherwise good circuits with poor implementation, etc. Such changes often raise the price significantly due to either cost of the components or just the fact that the device is "audiophile" category now.

None of these approaches merit you handing the company money. You are getting less and many times paying more.

SINAD therefore is a proxy and litmus test for design excellence. To the extent the designer worked hard to measure and optimize the design and knows what he is doing, then not only SINAD at 1 kHz is great, but likely elsewhere. I don't know of a single audio device with SINAD > 110 dB that has not benefited this way.

Remember that to get the highest SINAD, you must conquer noise. You are not only fighting the noise that your product produces, but also the noise in the measurement system! Noise is absolutely audible and simple depends on volume you play. Passive speakers are noiseless as are headphones. So they are not the limiting factor.

The only reason to ditch SINAD then is to want to buy blind and rewarding companies who don't do much to deliver excellence to you.

 

[Colonel7]

yet, most audiophiles say that it sounds awesome,

Do they now? Statements like these should make you check your wallet bc someone else has their hand in it!

 

[pma]

What if? What if NAD and PS Audio engineers deliberately added some odd and even harmonics, with specific magnitude ratios, so the precepted sound to the human ear will be optimized to have these audiophile characteristics, such as “warmth” and “depth”, which cannot be captured by the Audio precision analyzer? Maybe they did some research and came up with this secret recipe for an harmonic contamination pattern that does magic to the sound perception, but results in a cr@ppy SINAD?


What if we are all wrong here? Is it time to shut down ASR and lock Amir up?

Did anyone research this? Is there an AES article which correlates harmonic contaminations to sound perception?

Sorry, this is similar to reading fake news. Distortion does not bring any "optimization" to human perception of reproduced music. Audibility of different distortion spectrum profiles has been investigated and it is well known that lower order harmonics are less audible than higher order harmonics. And masking curves show it with respect to frequency and level.
And yes, there have been research studies on audibility of distortion profiles

https://www.researchgate.net/public...tortion_Profiles_in_Relation_to_Audio_Quality