The Audio Science Review (ASR) approach to reviewing wines.

@samureyex 

What has the better predictive value for the sound of a DAC: SINAD or price? I go with SINAD. Not perfect, but better. Scientifically expressed as higher factor loading in a PCA.

Why are expensive DACs generally reported to sound better? Because buyer would otherwise have to admit of having wasted money. Pure psychology.

I just bought a Topping D50III ($250) based on ASR assessment. If memory serves me well, it measured SINAD 123 vs. Molamola at 124. So the Topping appears to be good bang for buck, particularly for my little secondary desktop system. As an aside, the D50III gets some recommendations by audiophiles. I compared it to my HoloAudio May L2 ($5K; not in ASR listing) in my main system. Plugged both into Roon server, volume matched, linked the two sources in Roon, plugged both into preamp, so can switch in an instant. Listened to a bunch of different tracks and music styles from EBM (attack/punch) to indie pop (soundstage) to chamber (decay). On some I liked the May better, on some the D50III. Essentially, overly analytical recordings sounded better on the more laid back May, whereas more relaxed recordings sounded better on the D50III. The differences were very subtle, much less than different tubes, headphones, or cartridges. Bad recordings were equally unmasked by both. So a $250 DAC is not clearly worse than a $5K DAC. Price difference: 20x. Shockingly (not) it depends a lot on the recording.

ASR serves as a good buyers guide for those of us who rather listen to music than to gear. Not the only one, but a good source.

Did I waste $5K on the May? Arguably yes, though the D50III did not exist when I bought the May. Am I replacing the May with another D50III? No. I already have the May and it looks nice, but admittedly does not sound clearly better. I still have to compare the May to the D50 [OG] with SINAD of ~90

Re cables, again, you do not hear capacitance or resistance or inductance, just as lille as you hear color. As pointed out by others here, go start a double blind testing series with listeners, and report your findings. If the EFFECT of cap/ind/rest can be heard in sound waves, that will be most fascinating. A possible EFFECT of cap/ind/rest is frequency response, noise etc. and that is precisely what ASR measures.

I find ASR a refreshing, fact-based antidote to the florid nonsense hyperbole in traditional audio reviews.

Take Fremer who argued in a TT review in TAS last year that one arc second is audible with respect to wow&flutter. That is physics nonsense*. And the editor didn't catch that either. So I cancelled my TAS subscription again.

I also appreciate Amir's firm grasp of engineering, for instance that a USB reclocker is pointless (one of the latest videos). Plus emphasis of psychoacoustics, e.g. second one is better, as demonstrated by listening test of the USB reclocker, where it sounded better after the $4K reclocker was taken out. Plus the emphasis that objective testing of subjective listening test IS possible.

* for those who don't see it immediately, one arc second is an angular measurement, whereas speed stability if expressed as change of velocity over time (dv/dt). Even as a marine biologist I notice this. For fun, one can calculate the acceleration required to make audible changes in speed over 1 arc second, taking threshold of hearing for frequency (~2–3¢ on a good day) and time changes into account, and then it becomes even more ridiculous.

@knownothing Science is about looking at data and making the best possible inference with the least ad hoc assumption (Parsimony, Mach's Principle of Economy, Ockham's Razor). It needs not to be predictive or testing as is shown in e.g. medical diagnostics using adductive inference (see Josephson & Josephson, 1996: Abductive Inference: Computation, Philosophy, Technology). Popperian hypothetico-deductivism is not the only scientific inference. In fact, it is rather limiting, one could even say boring, as it is not ampliative. Popperian approach cannot find anything new, it can only reject something that is already there. I find and describe new species and that is certainly science and definitely not Popperian.

Amir's approach is certainly more scientific than TAS or SPh etc., which is nothing but fanciful storytelling.

I have watched quite a few of Amir's videos, and have also corresponded with him briefly over email. He is a no-nonsense scientist, like myself. One should not mistake clear articulated words as a sign of arrogance. I find him actually quite deferential.

@samureyex Re cables, the biggest difference should be from bargain basement to something better. I have tried that with speaker cables and interconnects, and have not found any difference between a Cardas 101 AWG 14 cable and a PearlAcoustics AWG 5-6 cable. Hooked up at same time on A-B speaker outs, so could switch in an instant. Same with interconnects. Re system, see my virtual system. Based on my personal listening experience, I cannot hear any difference between any cable I ever tried. This fits well with Amir's measurements, as well as Gene with Audioholics for that matter. 

@samureyex I agree with you that there are differences. The question is, do they matter? Are they audible? In any scientific test, the null hypothesis is no difference. So the burden or proof is with those who think there is a difference. If you think there is something to cables that is not measurable, then show it with a controlled double blind test of listeners.My preliminary data with vastly different cables and one listener is that there is no difference. Doing a pilot study with something that should show something is the typical way to start a project. If AWG14 vs 5–6 does not show a difference, that is a good indication that there will be none with other cables either, at least for this listener (inductive reasoning).

@knownothing Re the name, if that is your hangup. As I said ASR is *more* scientific than TAS and SPh etc. combined. ASR certainly provides novel data with measurements that are objective and repeatable. Re NSF, a good chunk of science is done without grant support. Neither of my two NSF grants had anything to do with hypothesis testing (one MRI, one TCN, both in DEB).

@samureyex @jrareform 

LOL! ASR s faulted for evaluating equipment based on measurements (which are actually relevant to audio such as frequency response, distortion), and then you adduce measurements that are INaudible (capacitance, resistance, maybe also color?) to support your claim of differences? LOL!!! Can you tell from listening that cable has 10 µOhm/m vs 20 µOhm/m? (no idea about actual values, as they are irrelevant to audibility). 

The real question is, are those different values audible, i.e. human perceptible? ASR makes excellent case that they are not, based on measurements that are RELEVANT to audio. That is why Amir makes frequent reference to threshold of hearing to put measurements into context of human experience.

Yes, Amir uses equipment that can measure differences way below the threshold of hearing, which is good scientific practice. Make sure you can possibly show differences relevant to the question at hand above noise floor of measuring equipment. I use scanning electron microscopy to look at 1–10 µm structures, although light microscopy could theoretically resolve down to 250 nm; in practice less due to diffraction, but that's a different story.

Re burden of proof, of course, nobody HAS to do anything. This is in context of scientific hypothesis testing, and there H0 is always no difference. Without having to show anything, scientifically there is no difference. QED.