Of Coarse Accurate! A Reviewer's View of the Audio Ideal - INNERFIDELITY

http://www.innerfidelity.com/content/coarse-accurate-reviewers-view-audio-ideal



I took some time to read through Tyll's new article this morning (BTW keep up the great work @Tyll Hertsens ). There's certainly a lot to digest, and since I'd rather walk into a gas chamber than an IF comment section, I figured I'd bring it up for discussion here.

 

I thought it was a really well put together piece and found myself nodding along in agreement to much of what Tyll said.

I completely agree that musical enjoyment is what everyone should really be after, but musical enjoyment is not a design goal from an engineering perspective. The people who are making and designing the products we all use need to have certain quantifiable and measurable criteria that they set for themselves. It not only makes the design and implementation process much faster but it also lets you confirm if the design goals have been met or not. If you simply set out with the intent to have a product that sounds good how do you know when you are there? Even if you say that your amplifier should have "fast" step response for transients it is still not clear what that means exactly. If you say you the amplifier must have a rise time of 2.5 μs you now have a good design criteria that you can validate. But why 2.5 μs? This is where the research for the link between subjective experiences and objective measurements come in. If you determine that there is some threshold for transient response at which the human ear can no longer tell the difference this might be a good objective. Or possibly there is some threshold below which imaging becomes clearly defined for the listener, this is also a good reason for an objective. Going for better numbers for the sake of better numbers though often can lead you down the wrong path.

The question then becomes why should the consumer be concerned with these metrics by which the designers are working to create their product. The answer for me seems to be that in a world with unlimited time and money you wouldn't need to be concerned with any of these you could simply purchase and trial all of the products and decide which one sounds best to you, however in the real world this would be very costly, time consuming, and tedious. Therefore these "objective" metrics can allow you to get an idea of the performance of the product before you buy it or listen to it.

An example from another world would be if you were buying a truck to tow large objects around, you wouldn't simply blindly buy any truck if you needed a 3 ton towing capacity. You could end up with a truck that only has a 1.5 ton tow capacity and obviously the performance would not meet your needs. The problem in audio is not only do we not necessarily have a way to measure the "tow capacity" of our equipment but in many cases we aren't really sure what aspects of the performance we should be looking at. Obviously choosing the fastest truck would not be the correct metric to look at in this case.

 

Read this late last evening while trying to go to sleep. A line from Wilde's "The Critic As Artist" came to mind,

Now, most of Wilde's aesthetic philosophy wouldn't mesh well with Tyll's (his Hegelian emphasis on Weltgeist, or World-Spirit; the subordinate relationship of the ethical to the aesthetic, etc. etc.), I think that this line is central to what Tyll's doing here. He's encountering a fundamental antinomy - an irreconcilable contradiction between two ostensibly true statements - in dealing with "accuracy" in audio reproduction:

1. When looking at studies done in the speaker realm, listeners tend to prefer tonally "accurate" sound over sound with gross tonal irregularity;
2. Because the multiple levels of mediation involved in the recording and production of audio, it is impossible to know with certainty what "accuracy" even is ("we have no way of knowing what the recording sounded like in the studio.")

This is a problem for a reviewer - how can we say that people prefer "accuracy" if mediated audio is only an index of the original event filtered through process and technology? Recorded audio is always a surrogate or proxy, a stand in for the original filtered through technology and recording process. Born digital audio takes that even one step further: it's a simulacrum, a copy without an original. There isn't even a "performance" there.

How can a subjective reviewer make claims to evaluation or judgement, which are both fundamentally related on some level to a claim to knowledge, with this level of central contradiction? Since the ethical responsibility of the reviewer is at the center of Tyll's thought process (through Toole's "responsibility" assertion), the correlative is: how can an audio reviewer ethically inform their readership about the sound quality of the equipment that they're using on their own via subjective claims? Answering this question could lead in a number of possible directions.

Do we discard any relationship to accuracy or faithfulness, seeing them as impossible claims towards a unknowable truth-of-audio, and focus on hedonistic audio pleasure instead? Is it just the pleasure principle, the priority of taste preference, that matters in the subjective experience of audio? Tyll criticizes Gutty for on some level doing just this thing. (And dismantles his use of the ER4-PT as evidence in the process. Just beautiful.)

Should we rely purely on the correlation of measurement with subjective impression to make any claims to knowledge? Nope again, because as Atkinson puts it, "All measurements tell lies."

The elements of Tyll's "reasonable method" seems to me to make sense. As I see them:

1. Make only claims to limited, subjective knowledge;
2. Measure what you can, but present measurements as helpful data and not absolute truth;
3. Use a standardized, set number of evaluative recordings - music or pink noise tracks - designed to test a certain, specific aspect of the playback experience. Know these recordings well;
4. Make a good faith effort to describe what you hear - and if you have to use "accurate", know that you're talking about "subjective accuracy" which correlates with an referent that may or may not be universally generalizable;
5. Beware the Jabberwock, my son! (In this case, Jabberwock = the mapping of your own emotional state on to a particular listening experience or recording);

They're pragmatic - and they allow you to navigate the space between and around the fundamental antinomy, enabling the reviewer to make reasonable, not-inflated claims to knowledge while simultaneously respecting the difficulty of the task and the incompleteness of a single person's perspective. They allow you to act responsibly and ethically in the face of the unknowable. Nice work getting that out, Tyll.

Post-script: One thing should be painfully clear from this piece: f**k "transparency." If "accuracy" is an incomplete, elusive, and possibly-unknowable quantity then "transparency" is simply the index of the reviewer's own ability to imagine an audio performance. No piece of equipment is a time machine - you need a police box, telephone booth, or DeLorean for that.

 

Thank you, guys, great comments.

 

I think a central point in Tyll's article is the idea that, at least in the case of educated ears, a subjectively "better" or "more accurate" sound sound does correlate to something most listeners agree is better, and also measures closer to something called "neutral".

Put another way, there is a lot of second guessing among people about whether something that sounds subjectively better is actually better in a real sense, or just different. Tyll makes the point that most listeners find greater listening pleasure as the reproduced sounds moves towards "neutral", and that educated ears do indeed favor the same "neutralish" sound. ie We can quit worrying, subjectively better is better (again, for ears that have some experience).

@Tyll Hertsens correct me if I didn't catch some of the nuance there.

 

Yes, that's pretty good. Measurements are a significantly incomplete, but substantially reliable. Subjective experiences are complete--in the sense that the listening activity closes the loop in the "artist expression to listener experience"---but they're unreliable in term of being transferable from person to person.

None the less, in the end, it is about the listening experience. That's the comprehensively meaningful event.