This is a rehash of the same argument. If you go low enough in numbers, you can hear it. Sure. No one disputes that. But no one has defined a threshold where it gets a pass mark because individual hearing differences do not allow for such a clean boundary. And as you go up, somewhere it becomes inaudible to anybody. Nobody knows what that minimum number is either but one can always take a safe bet by picking a number. In addition, there are multiple components going into that number. It could be audible in one component but not the other. But two similar composite numbers may be at opposite ends in those components and therefore one may be fine and the other not.
It is in this sense, the correlation between measurement and audibility is neither well defined nor correlated.
The best you can say is that at a suitably high number, the distortion and noise won’t matter to anybody and therefore there are two buckets - pass or fail with respect to that number. That is exactly why the SINAD ranking is just a measurement ranking not an audibility ranking.
When people look at Okto’s numbers and ask “does it sound as good it measures?”, then you know there is a problem with misunderstood correlation with measurement and people’s need for good audible sound rather than measurement.
Good science would require that you establish that it has indeed exceeded the audibility threshold before opining on it. Not “may have”. But since that audibility threshold itself is ill-defined the above is not a scientific statement. Confirming it with listening tests would require (1) well documented testing methodology that can be analyzed and (2) either a causal explanation or at least correlations between such audio deficiencies (clarity, stage depth, whatever you want to pick) and the actual thing measured at that level has been conclusively established via a study of units with similar measurements and units that are better in that measurement.
Anyone with some knowledge or experience in science experimentation would understand the above.
I am sympathetic to limited resources in doing so, but that is no excuse for making definitive claims.
Good, that is a start. From now on, I expect people would not use that table to make any claims about better or worse in audibility. Perhaps a revisit of the AVR threads would be in order with that in mind.
And if they do make such a determination, they would justify it with more than where it falls in the SINAD table.
This is a very poor use of the word correlation. Positive correlation would imply better sound quality with better numbers and negative correlation would imply the opposite. When no such established correlations exist it is said to have no correlation (zero correlation would be more stringent requiring showing that the relationship is statistically random).
The best answer you can provide is that below a certain threshold that is hard to pinpoint and varies, the measurements may indicate artifacts that are audible with no measure of how audible. Correlation is not the right word for it.
I think you're stirring the pot a bit too hard.
You appear to be basing your thesis on one measure of one metric. Just for indulgence sake, let's address this.
It would be fallacious to pick the highest SINAD level at which things become inaudible. People have different hearing. They listen to music at different level. They do so in different spaces, using more than a DAC.
BUT, we can say with the weight of medical science behind us that human hearing has a limit of -118 dB. Does that apply to very single person? Who can say. But it has been accepted in science which has undergone a very high degree of peer review. So in the absence of better info, it's a reasonable place to start.
Now this raises a legitimate discussion on whether or not noise and distortion is detectable by human hearing in normal listening conditions. Normal is a broad and undefined term, so it is wise to use a standard definition. That is, whether you like it or not is 85 dB and is sometimes expressed as 0 dB because it serves as a reference point. The 0 dB reference point isn't arbitrary... it is the threshold between the highest spl of audible broadband sound that people can listen continuously to without suffering some form of permanent hearing damage.
If I understand this correctly, an unimpaired human should be able detect sound at -118 dB from the reference point. How much distortion? That is another matter entirely because it is not
simply an issue of spl but of frequency, harmonics, and percentages. It is highly dependent on space too.
How much does masking play a role? What about reinforcement? Depends on other gear being used, none of which Amir can "see" because he doesn't dictate the terms under which everyone on planet earth listens. In other words, it is an uncontrollable variable so shouldn't play any definitive role in
his assessment. This is something I will get back to at the end of my post.
Even considering all of this, medical science tells us that sounds occurring at -119 dB are technically inaudible by the vast majority human beings. This is something eminently useful to consider, and a point I will use at the end of this post.
Psychoacoustics holds different views and, based on your posts, it appears you are a fan. But psychoacoustics remains a developing field. So much so that there is considerable debate between psychoacousticians on the levels at which sound can be perceived (as opposed to heard, because psycoacousticians have been unable to demonstrate, in any scientifically accepted manner, that human hearing extends beyond -118 dB). In other words, psychoacusticians are not in agreement on the fundamentals of the field.
You appear fixated on SINAD. This isn't because you don't understand what I've written above. I think you have an excellent grasp of audio. Instead, I see your quest for a single answer as an ideological pursuit in which your goal is to redefine the significance of established SINAD thresholds. One of your methods of discussion is symantics, which I can appreciate were it not for the fact that all such discussions usually start nowhere and end nowhere (that is their nature, after all).
So I think you might be missing (or dismissing) the essence of standards. They are not absolutes but benchmarks. They confer no meaning on their own but can be used to derive great meaning when combined with other data. Because that other data can affect the outcome greatly, it can only be resolved on a case by case basis. When we go down this path, the symanticist then says "yeah but", leading the discussion on another tangent leading to nowhere.
This doesn't achieve anything other than to serve as a get out of jail free card to expensive, audiophile shite that trades opportunities for high performance for the certainty of high profit (assuming there's any engineering competence involved in the endeavour).
The great thing is that Amir makes enough of his measurements (of different aspects of performance) available. This allows users to independently draw their own conclusions, using (among other things) the variables of
their listening conditions and capacities.
So nobody is holding a gun to your head, or mine, forcing us to accept Amir's net assessment.