Daniel Johnston
Headphoneus Supremus
Overall I mostly agree with your observations and reasoning. Thank you for stating your views and points eloquently and non condescending.“Stupid” is generally a bit harsh in the context of misunderstanding a study, depending on nature of the misunderstanding but then I personally very rarely use it and would never use it in the given example. I was really just quoting your use of it and along the lines of “even very smart people can occasionally do something stupid or fall into a stupid trap”.
Hmmm, not sure I can agree with that. You could *maybe* argue it’s partly true if one stretches the interpretation of the word “anecdotal” to its absolute limit but that would be “anecdotal at worst”, rather than “at best”. Clearly there’s a different level of scientific reliability/validity between a controlled study and an audiophile doing a sighted, uncontrolled test and reporting “even his wife could hear the difference”. Certainly a scientific study never provides absolute proof, however, in the context of a body of evidence (and the lack of reliable opposing evidence), it can provide “near absolute certainty”. For example, the audibility of HD vs 16/44.1; There have been numerous studies on this, of varying reliability/validity, indicating this difference is not audible but taken in context of “the body of evidence” we have near absolute certainty. EG. The only evidence against this conclusion is from purely anecdotal, uncontrolled tests plus a clearly manipulated meta-study. And, despite the obvious significant financial incentive, after quarter of a century or so there’s still no reliable evidence against this conclusion. In addition, everything we know about hearing thresholds dictates that 16/44.1 should be more than enough anyway. Taken with some other factors I haven’t mentioned there’s no rational reason to doubt the conclusion, although this doesn’t rule out the extremely remote possibility that it could never be disproved but it would require some extremely robust evidence to even start down that road.
I’m not sure that’s exactly what he was saying. Certainly the earnest scientific study of audio perception started in the late 1890’s and those findings are just as true/valid today but that doesn’t rule out the possibility of further growth/understanding. An analogy to hopefully explain what I mean: “1+1=2” is very likely one of the oldest scientific/mathematical findings in existence, probably older than 8,000 years. Obviously science/mathematics has moved on massively since the Stone Age but 1+1=2 is still absolutely true.
That’s not quite true. As a general rule with musical material, audibility of jitter is between 200-500 ns, although with certain pieces (with specific, rare properties) it has been demonstrated that a few individuals can differentiate down to just under 30 ns. Using test signals (not music), detectability has been demonstrated down to around 3 ns. The Benjamin/Gannon paper (“Theoretical and Audible Effects of Jitter on Digital Audio Quality”) is one of the best/most comprehensive sources.
Hmmm, the average threshold for an adult is about 16kHz. Even older teenagers can rarely hear above 17kHz, so 20kHz is already a pathological exception. Plus, the rare exceptions of hearing so high was with test signals (pure tones) at very high playback levels. In the presence of multiple tones this discrimination ability is significantly reduced. And lastly, with musical signals the >20kHz content is virtually always far lower in level than the content below 20kHz. For example, with a violin, only 4% of the energy it produces is above 20kHz. Taken together, with the other factors (air absorption of higher freqs for example), as a whole body of evidence, it does pretty much rule out that an adult can hear >20kHz when listening to the reproduction of music recordings.
See my 1+1=2 analogy above. Is that nothing more than historical reference or does 1+1 still equal 2? I’m sure some audio perception studies from those times have been superseded but many were surprisingly accurate considering the level of reproduction technology available at the time.
*Why* we hear what we hear is still not fully understood in some cases but what we hear (without the “why”), in terms of most of the thresholds, is fully understood.
It depends of course on exactly what/which claims you’re talking about. Certainly some/many of them should absolutely be dismissed. For example, I recall Rob Watts claiming to hear distortions at -300dB, which clearly is ludicrous. One ludicrous claim is enough to warrant a great deal of skepticism of any of his other claims but unfortunately he hasn’t restricted himself to only one ludicrous claim. Of course this doesn’t rule out the possibility that he/they might make a claim that is actually true but if he/they make claims contrary to “established norms” then dismissal is a reasonable default starting point (which *might* in some rare circumstances warrant further enquiry)!
G
My sticking point is what studies are “1+1=2” and what studies could do with an update using modern equipment. I don’t have an answer. I’m still exploring and trying to understand what my ears and brain tell me vs measurements and other objective data.
To clarify, I’m only going on study design. In my profession, a randomized, double blind, demographic matched, case controlled study that is confirmed and reviewed by qualified and credentialed peers is as close to scientific fact as we can get. Even then, when parameters change or technology changes, these “facts” can be disproved or changed. I’ve seen it happen multiple times.
Some science is irrefutable fact (earth is round, 1+1=2, etc…), some is more pliable. Anything dealing with the brain and human senses is particularly difficult to randomize or control for bias or other factors. For example, if a study subject is stressed or angry the day of the study, this could negatively affect their perceptions.
Lastly, anecdotal evidence isn’t bad, nor weak by any means. It’s evidence that can’t necessarily be exactly reproduced or applicable to a different population, person, or test group. It’s often evidence like this that leads to larger, more powerful studies. I don’t mean to suggest that an an audiophile saying they hear a difference in a sighted, biased ABX vs one who set up a more scientifically sound level matched, blind test are equivalent.
Last edited: