Did Amir Change Your Mind About Anything?

@mahgister I just noticed this thread from my weekly Audiogon roundup. I am actually familiar with Gibson and his ecological approach from grad work in cognitive science. I took a moment to check out some of the papers related to your unnecessarily long and murky posts here, as well.

I don't think ecological perception critiques about spectral analysis are relevant to musical reproduction devices. They certainly are interesting in terms of explaining human listening experiences where expectations and environmental affordances certainly play a part in how the brain perceives the sounds emanating from a device. But if the goal is just to successfully reproduce audio with minimum noise and distortion, and with maximum fidelity to the original recording, I see nothing to suggest that following the guidance of sampling theory will not result in exactly the kinds of "transparent" or "uncolored" devices that are available today. Gibson then gets to critique how the human hears/understands the purity of the emerging sounds, and reconciles them with all the affordances of space, room, materials, mood, and much else.

There are edge cases where general psycho-acoustics can be influential, like using compression techniques that de-emphasize parts of the spectrum. We would prefer to de-emphasize only where the results have low impact on human listening, for instance. Phantom center images, Dolby Atmos, etc. certainly are another area where there are great research opportunities, too, for the ecological perception-focused researcher.

In any case, I have learned a great deal on ASR and recommend it highly. It provides an excellent counterpoint to vague assertions and hush-voiced listening reviews. Most interesting for me, however, is the internet culture role of how it is deconstructing the faith aspects of the audiophile subculture. We see that playing out here!

@mahgister Well, I just skimmed that response a bit but, yes, it certainly is possible to build audio reproduction systems that have colored sound that some people might prefer (tube and vinyl aficionados rejoice!), but in reality these were always stopgaps towards perfect fidelity due to the limitations of the devices. But some folks prefer them, so be it.

But with the advent of better technologies and the theories that guide their use, we can build remarkably transparent systems for low cost these days, so the limitations of those other approaches become more obvious. There is some research that shows that hypersonic sounds (>20kHz) might cause some brain activity changes, but it's unclear whether that is definitive for providing perceptually-relevant high-frequency components that would enhance our temporal experience of music, etc.

If you want to know what people want from speakers, you just need to study them and come up with a crowd-sourced preference curve, for instance. And lo, we have one.

But in any case, I remain perplexed what you hope to gain by pushing this line? If you have a desire to research how to color sounds to enhance audio enjoyment, please do the work. I will be interested (if it's well-written and coherent). Nothing you mention has any bearing on testing whether audio devices are high quality from the standard of low noise and distortion. It is orthogonal to those goals.

@mahgister Actually, I just read a bunch of van Mannen for fun. Luckily I have both a BSEE and MSEE in information theory and signal analysis, thus feel somewhat competent to comment a bit (though I prefer a much more reserved approach to science and engineering). A key point he mentions is that sinusoidal signal sweeps don't fully characterize the frequency response of, say, an amplifier, because of some theoretical requirements. Linearity is one of those requirements and there are all kinds of nonlinear things going on in real systems. Indeed, the effects of nonlinear transfer functions can be quite interesting and require very interesting mathematical tools. But, really, it's what we call distortion.

So, if an amplifier designer is trying to create a great amplifier, what should she do? She could test using sinusoids and try to reduce distortion and noise in her design or she could do....what...exactly? van Mannen has concerns about feedback topologies as well, but still, other than trading-off options, she still will want to test using the best tools she has in an effort to reduce noise and distortion.

@mahgister Fair enough. Having also worked on studies of human subjects in later careers in technology, I can assure you that subject reports are clouded by an enormous range of biases, expectations, timing effects, etc. Amir goes into the problem of performing even simple tests to distinguish systems and how challenging it is, so I have strong doubts about the subjectivist-style claims and how intellectually honest those folks are!

I just don't see any downside to leaning heavily into objective tests. I've only seen poorly-designed tests (often headphone tests are impacted by problems with the ear cups, for instance) that were problematic.

Anyway, I learned some new stuff today! Always a good day...