The Audio Science Review (ASR) approach to reviewing wines.
Imagine doing a wine review as follows - samples of wines are assessed by a reviewer who measures multiple variables including light transmission, specific gravity, residual sugar, salinity, boiling point etc. These tests are repeated while playing test tones through the samples at different frequencies.
The results are compiled and the winner selected based on those measurements and the reviewer concludes that the other wines can't possibly be as good based on their measured results.
At no point does the reviewer assess the bouquet of the wine nor taste it. He relies on the science of measured results and not the decidedly unscientific subjective experience of smell and taste.
That is the ASR approach to audio - drinking Kool Aid, not wine.
+1 @2psyop Not sure why these ASR sucks threads keep coming up when half of the posters say they don't read ASR anyway. It's just information, and no more or less significant than written reviews without context or full of personal bias. Do your research, and then buy what you like. Do you buy wine based solely on a review? Reviews can spark interest, but at some point you need to open a bottle and drink it yourself.
|
As an electrical engineer who's spent close to 30 years designing and building hi-fi electronics and loudspeakers, I have quite a few things to say on this topic... I'll try and be concise. Objective measurements are great tools insofar as the results are understood and interpreted properly. This is where the current debate seems to run into trouble. Take SINAD (aka THD+N) for example. There seems to be a monomaniacal over-emphasis on this metric as an end-all-be-all measurement which somehow dominates the subjective performance of a piece of equipment over most other aspects of performance. Since we're doing analogies: SINAD is like a ruler. It measures one, and only one, parameter of the DUT. Just as a ruler can measure length but can do nothing to measure density, hardness, color, IR emissivity, chemical reactivity, unladen airspeed, or any of a thousand other possibly relevant properties of an object, SINAD / THD+N tells you nothing about most aspects of frequency domain or time domain performance of the equipment, both of which are extremely relevant to how it will subjectively perform. I would even argue that THD in the context of electronics is increasingly irrelevant, given how low distortion is in most modern designs. Turn to another famous objectivist like Ethan Winer, and you'll find excellent demonstrations of the audibility of THD. His demos area easy to find, and the shocking conclusion is that anything below ~1% THD is (or can be) essentially inaudible. Indeed the conclusion of the pioneering scientists in this field (close to a century ago) was that 0.1% THD represented a reliable threshold of subjective perception. And your loudspeakers are unlikely to do better than about 1% THD as well. So why chase after 0.0002% THD in a DAC or amplifier? I've built, lived with and loved tube amplifiers with rather embarrassing distortion figures compared to the modern benchmark. I've also built solid state amps following the guidance of famous objectivist Douglas Self, and while the result measured extremely well (~0.004% THD), the subjective listening experience of my earliest efforts was... disappointing to put it mildly. I kept using my tube amps while spending years trying to decipher how to build a better sounding solid state amp. So measurements have their sensible limits as well. It does no good to go overboard with a single specification. "But", you may quibble, "Amir at ASR has a whole suite of measurements intended to cover every aspect of good performance". While I commend Amir for his attempt to cover all the angles, IMHO his measurement selection still misses a number of relevant subtleties. Most notably the time domain performance of DACs and amplifiers, which are essentially not measured at all by instrumentation like the AP analyzers (they rely upon steady-state sine stimuli only). First-rate engineers like Bruno Putzeys and Julian Dunn have pointed out the importance of subtle things such as passband ripple in digital oversampling filters, and of reaching full stop band attenuation below the Nyquist frequency, yet these important aspects of DAC performance continue to be widely overlooked. Possibly because they are harder to measure and interpret the subjective effects of deficiencies? I don't know. But I do know this: a ruler (even a really really awesome one) is just not good enough. The human perception of sound is not well understood and even less well quantified, and there are many aspects of objective technical performance of audio equipment that we already know something about which are being overlooked. At the end of the day, it all boils down to this for me: Objective audio measurements must by definition be subservient to the Subjective outcome. If not, then we aren't talking about hi-fi anymore. As Nelson Pass so eloquently puts it: "We are in the business of entertainment, not dialysis." I am fairly sure that "the science" (or really: the science as practiced) will eventually catch up as better measurement techniques are created and gain widespread acceptance. It's just that at the moment, the audio industry as a whole has stagnated on a small set of decades-old methods. Part of this is a marketing problem too. The "objectivist" types are rightly critical of an unfortunate degree of snake-oil peddling and pseudo-scientific misdirection that goes on within marketing departments, and elsewhere under the guise of "subjectivism". Toss in a dose of Dunning-Kruger effect, and the usually deep innate desire of every audiophile and their ego to be an expert, and we have a different problem category, which the ASR types are doing their best to shut down. I applaud this aspect of their endeavor, but sometimes they take themselves too seriously. So take it from a die hard engineer and objectivist-subjectivist fence sitter: Trust your ears, and buy the gear that makes you happy. |
