1) You pointed out how a null test will demonstrate even the tiniest differences... I agree entirely, however, as I pointed out, if you null two of anything carefully enough, you always fine a little difference...
[1a] We always end up back at "now that we know exactly what the difference is we have to decide whether it counts or not". At which point someone chimes back in with that mystical knowledge that a certain null is "good enough we can act as if it's inaudible without bothering to test it". (I suspect you're right, and that some difference is "small enough to be inaudible", but I'm not quite sure exactly where that number lies... maybe we need to test it.).
2) I personally am willing to accept that the results from those particular tests were what you say. (But a lot of other folks on this forum seem to think that "all claims are anecdotal if we don't get the data that goes with them".)
And, yes, those results are interesting... especially that they show that, at least under those conditions, experience seems to be more significant that "raw hearing acuity".
[2a] (But we still don't actually know if the massive difference in the hearing acuity of a five year old would override that difference - until we test it.)
4) Since a single outlier renders you unable to make a generalization about "everyone" you have to try as hard as you can to find all the outliers. For example, there is a lot of incentive provided for fast human runners to come forward.The possibility of an Olympic gold medal, and a few $million in endorsements, makes it extremely likely that most fast runners HAVE been clocked. However, there is no such clear-cut incentive for people with exceptional hearing to come forward and be tested. Therefore, in practical terms, only a tiny percentage of every human on Earth has even had their hearing tested.
1. Firstly, if you "agree entirely" that a null test "will demonstrate even the tiniest differences" then your previous post was at best misleading: "
If you hear a difference, but can't seem to find a measurement that would account for it....- maybe you're simply not measuring the right thing ... - maybe you're not measuring it accurately" - Because a null test, one of the oldest and most commonly used tests, completely covers all these assertions! Secondly, it is NOT true that if you null two of anything you will always find a little difference. For example, if you null two identical digital audio files you will find absolutely no difference. Your assertion is true only for analogue signals/components, due to the random nature of thermal noise. However ...
1a. You appear to yet again be employing that same typical audiophile fallacy/falsehood. Maybe you personally haven't tested it and/or are not aware of the testing which has been done but it's a complete falsehood to state it doesn't exist just because you personally are ignorant of it. In actual fact the exact opposite is true, it's been tested exhaustively, over numerous decades by countless thousands of people! You say we should obviously include some common sense but then don't. For example, when I'm dialogue editing I will typically employ a null test 50 or more times a day. You think maybe that if the test reveals a small difference I just say to myself "oh well" and move on or do you think I test it to find out if it's audible and by how much? Actually, a lot of the time I do just move on, because after 20 odd years and god knows how many thousands of tests to determine if the difference is audible, I can often tell just from looking at a spectogram of the difference file. I still end up having to physically test a fair proportion of the time though, probably about a third of the time. And of course, I'm just one of many thousands of engineers who edit dialogue around the world. Furthermore and again, countless thousands of sound engineering students are taught how to conduct a null test, determine the significance of the result and demonstrate the ability to do this under examination conditions. Also of course, there is a wealth of scientific evidence to draw from. As simple and obvious example would be that if the difference file exhibits a small difference that's above 20kHz, then the science tells us it will be inaudible and this is just one of several similar examples.
2. But of course, we need to apply some common sense! Anecdotes ARE evidence, although typically the least reliable form of evidence but also, I don't recall reading a single published scientific hearing study that didn't have at least one flaw. Audiophiles will therefore often invent a completely fallacious equivalency; anecdotes are flawed and so are scientific studies, so I logically choose to believe the Head-fi poster who actually owns a Chord Hugo over a scientist who probably doesn't even know a Chord Hugo is, let alone what it sounds like! This is a complete fallacy though because omits common sense, the common sense that all flaws are not equivalent and neither therefore is all evidence. Even within anecdotal evidence one can apply common sense and give it more or less weight. For example, I would tend to give more weight to the anecdotal evidence from Bob Katz than to the average head-fi poster because I know that Katz has considerable knowledge/experience of controlled testing while the average Head-fi poster probably isn't even volume matching. Even so, as it's anecdotal evidence I still wouldn't just accept a Katz anecdote as fact, unless it was corroborated by more reliable evidence. It's up to the individual reader whether they choose to believe the test results I have reported, which is effectively anecdotal as I can't provide the actual data to corroborate it. However, common sense dictates that there must be considerable testing, hearing ability and listening skills are absolutely fundamental to a music/sound engineer and therefore it's common sense that students hearing is tested, that they're taught listening skills and those skills are tested/examined. Therefore, it's both factually incorrect and contrary to common sense to state that "we" don't test!
2a. There's an obvious problem with your statement. You assume that we just individually decide that a difference is inaudible and that's the end of it. Is it common sense to assume that music/sound engineers differ from other human beings in that we are all completely devoid of any curiosity? If all competently designed ADCs/DACs are transparent (have no audible differences) then no competent engineer would ever test them, however we do, regularly, because of curiosity. But how do you audibly test inaudible differences? Again, common sense, we magnify/amplify the differences. For example, it's standard procedure to loop a recording through an ADC/DAC ten times and audibly compare the result with the same 10x loopback recording through a different ADC/DAC. We've effectively magnified the artefacts of each of the ADC/DACs by 10 times and the differences are audible. Or, when testing different dither algorithms, we record the dither and then amplify it by 40dB or so and then we can audibly compare them. The obvious problem with your statement is therefore "
the massive difference in the hearing acuity of a five year old", there's no evidence to even hint at the possibility that five year olds have a massive difference in hearing acuity, the evidence suggests a difference of very roughly 10% and possibly as much as 20% compared to an adult but that's no where near the 1,000% or 10,000% which we commonly have to apply in order to discern differences!
4. Again, we need to apply some common sense! Hundreds of thousands of people, probably well in excess of a million, have had their hearing tested over the decades. Is it possible an outlier exists who has 10% more hearing acuity than anyone ever measured? Sure, somewhat unlikely but certainly possible. What about 10,000% more hearing acuity? No, that's not credible. The physiological structures in the ears aren't capable of it and anyone who had say 40dB more hearing sensitivity than me would be in constant pain, they wouldn't be able to function normally and they would have been discovered.
[1] And hearing a really loud noise that makes you jump once in a while isn't dangerous... and may add a lot to your sense of realism.
[2] And someone else may just occasionally "want to make their ears bleed" - even though they know it's probably not a great idea.
[3] (But, yes, if you want to claim that your system can reproduce a symphony orchestra accurately, then it should be able to recreate the original levels involved, including the cymbal crashes, and the cannons.)
[4] I should also point out that the idea that "a device only needs to be able to perform exactly as well as you need it to" is somewhat misleading.
[5] It's also worth noting that our perception of loudness varies considerably.
[5a] An average level of 80 dB is going to sound far louder to someone arriving home from their job as a librarian... Than it is so someone arriving home from their job repairing jet engines...
1. That depends. A really loud noise that makes you jump doesn't actually need to be really loud, the human perception of loudness is relative. See #5 below.
2. The problem with that of course is that who in their right mind would want to make their ears bleed? This appears to be a quite common view amongst Americans, that they should be free to make their ears bleed if they want or for that matter, own a gun and shoot themselves in the head. The problem is that in reality the majority of people who do make their ears bleed will either not be in their right mind, in which case you are enabling those with mental health issues to self-harm or will be children and adults who do so by accident. For this reason the EU has strict legal limits on the max level of headphone output on mobile devices, as several studies indicated young people's hearing was being damaged and incidentally, we also have strict laws on gun ownership. Also, if the level has to be raised to an "ear bleeding" level in order for some artefact to become audible, then that artefact is inaudible.
3. Yes, but again, how loud is a cymbal crash? At 2" probably 120dB or more but at the ideal audience position, probably lower than 90dB. BTW, cannons are not a standard member of a symphony orchestra!
4. Applying common sense, the response is that it depends. Sure, many/most amps become increasingly non-linear and noisy beyond around 80% of max amplification and I would always advise an amp powerful enough to allow at least 20% headroom. A DAC on the other hand does not require headroom.
5. Yes, to start with, it's relative. An 80dB sound will sound very loud if it's preceded by say 40dB sound but if it's preceded by 75dB sound it won't. This fact of perception has been routinely employed in film sound for many decades, if fact audiences are commonly so expectant of a subsequent loud noise when the level gets lower, that we can use that expectancy to further surprise them.
5a. Our hearing can adapt quite quickly, just a few seconds in the case of loudness within safe levels but up to several hours if levels have been extreme. In other words, an average level of 80dB is going to sound about the same to a librarian as a jet engine repairer as their hearing would be adapted to the level of car/traffic noise on their way home.
I think Gregorio might be basing his loudness estimate on standard miking positions for recording in an empty house, or perhaps short impulses of less than a second. That isn't the same as a loud listening level for commercially recorded music.
You'd have noticed that I gave very different and higher figures for being close to the orchestra than for being in the audience, the close figures are based on mic'ing positions but the figures for being in the audience were not, they were based on my experience as a live sound engineer and with audiences being present. However, I'm typically basing my peak levels on a window of 0.3 secs, a peak level using a slow response would measure somewhat lower. Also, for safety sake, I'm basing my peak level of around 90dB on the most extreme examples. Although very loud for it's day and still very loud compared to say a Mozart symphony, Wagner's ring isn't as loud as it gets. Later pieces/symphonies can get louder still, Mahler being a good example but Holst, Stravinsky, R. Strauss and various others. For the vast majority of audience members, for the vast majority of performances, in the vast majority of concert venues and using a slow response to measure peaks, 80dB or so would likely be about the maximum.
The noise floor for the room measures about 32 dB.
BTW, typical SPL meters are extremely poor at measuring low levels/noise floors. Typically they're only accurate for levels of around 60dBSPL and higher. Unless you have a meter specifically designed for low levels and/or a very expensive one (thousands of dollars) with particularly low self-noise, your measurement of 32dB is likely to be out by a significant amount.
G