The deaf leading the blind? A piece by Henning Møller (B&K)

[tuga]

Radical "Objectivists" are afraid of listening as much as alt-"Subjectivists" hate measurements.
Are the deaf leading the blind?

There are examples where problems were identified through listening assessment and then people tried to develop ways of measuring those problems (i.e. TIM).
The subject is currently being discussed in the following topic where Jussi "Miska" Laako (HQPlayer) describes how he is triyng to nail down an audible problem in the performance of ESS Sabre D/A chips that doesn't show in a "traditional" set of measurements:

https://audiophilestyle.com/forums/...ements-correlate-with-subjective-impressions/

I think that radical "Objectivists" that do not value critical listening are nearly as deluded as ears-only alt-"Subjectivists".
Critical listening and measurements are complementary tools.


This interesting old piece presented by Henning Møller (Brüel & Kjær) at the 59th AES Convention in 1978 discusses listening and measurement correlation:

Abstract
Audio is easily and meaningfully perceived by the "global" subjective human mind - and comprehended simultaneously. A similar "meaning" can be obtained in the objective world of measurements if - as in the human mind - a reasonable amount of "local" objective measurements are simultaneously considered and weighted. No single measurement is sufficient. Today there are already six "measuring domains" that strongly correlate to the subjective perception of Audio. These will be discussed in the paper.

 

[tuga]

I've just bumped into this follow-up on Møller piece by Tim Gladwin (Harman) which ties it up to the Harman research and CTA/CEA2034:

Chasing the Elusive “Good Sound” Meter for Loudspeakers

http://www.loudspeakerindustrysourc...ve-aaagood-soundaaa-meter-for-loudspeakers-32

 

[JMC]

Instruments are tools for measurement. Ears are tools for measurement. One is realiable and repeatable, one is not. Questioning objective, instrumented measurement only makes sense if the problem is so complex that you cannot make meaningful measurements of all important parameters. I do not believe this is the case

If you look at closely related fields and a different measurement sense i.e. your eyes, you see there is less (or no) confusion or "controversy". Things that measure objectively better are considered by all to be better. HIgh end audio relies almost entirely on two things, suggestibility and the unreliable measurement of our ears.

 

[q3cpma]

Instruments are tools for measurement. Ears are tools for measurement. One is realiable and repeatable, one is not. Questioning objective, instrumented measurement only makes sense if the problem is so complex that you cannot make meaningful measurements of all important parameters. I do not believe this is the case

If you look at closely related fields and a different measurement sense i.e. your eyes, you see there is less (or no) confusion or "controversy". Things that measure objectively better are considered by all to be better. HIgh end audio relies almost entirely on two things, suggestibility and the unreliable measurement of our ears.

To add to this, when the "measurements aren't enough or even relevant", we know it, we don't guess it. Look at the video field, basically anyone with a little interest knows that PSNR and SSIM (including their more advanced versions like PSNR-HVS and MS-SSIM) don't correlate well enough with MOS; VMAF is the only thing that's not completely stupid. The best encoder (not format) we have, x264, got here by more or less completely ignoring measurements and focusing on psychovisuals, same for LAME in audio.
On the other hand, Toole proved that the correlation between FR flatness and directivity/power response smoothness is high enough that everything else can be ignored unless it's botched. Some holes remain like a way to perceptually measure distorsion (i.e. not THD+IMD) and the absolute audibility of group delay and spectral decay.

 

[tuga]

Ears are tools for measurement.

Ears are not tools for measurement.

 

[tuga]

If you look at closely related fields and a different measurement sense i.e. your eyes, you see there is less (or no) confusion or "controversy". Things that measure objectively better are considered by all to be better.

Indeed.
I always use the VHS vs DVD analogy; no one watches VHS tapes anymore but people are still listening to mp3s or even cassettes.

HIgh end audio relies almost entirely on two things, suggestibility and the unreliable measurement of our ears.

High end is a marketing category. High-end real-estate, high-end cars, high-end hotels... It's a synonym for upmarket or deluxe.

 

[tuga]

Some holes remain like a way to perceptually measure distorsion (i.e. not THD+IMD) and the absolute audibility of group delay and spectral decay.

Thanks for mentioning it.

 

[JMC]

Ears are not tools for measurement.

Explain what they do then...

 

[q3cpma]

Thanks for mentioning it.

Still, don't forget they are tiny holes and can't be used by audiofools to justify their overpriced shiny crap.

 

[tuga]

Explain what they do then...

The Encyclopædia Britannica describes measurement as the process of associating numbers with physical quantities and phenomena.

To the best of my knowledge our ears can be used to enjoy music, to taste a sonic presentation or (for observational purposes) to identify, but not quantify, shortcomings.

 

[mansr]

The subject is currently being discussed in the following topic where Jussi "Miska" Laako (HQPlayer) describes how he is triyng to nail down an audible problem in the performance of ESS Sabre D/A chips that doesn't show in a "traditional" set of measurements:

Let me guess, his solution is "DSD" and everybody should buy his software.

 

[tuga]

Still, don't forget they are tiny holes and can't be used by audiofools to justify their overpriced shiny crap.

Maybe we could raise above the "audiophool" rhetoric and focus on performance.

Imagine that an audio equipment is an F1 car. The selection of tire compound hardness maybe even tire pressure, an adjustment to the suspension, different settings of wing geometry may produce tiny gains of a few hundredths of a second but when combined could give you a full tenth. And that may be enough for pole-position.

If you've had hands-on or listener experience with modified/optimised equipment you will know that simple improvements in the PSU or grounding can have consequential, audible as well as measureable, impact.

 

[tuga]

Let me guess, his solution is "DSD" and everybody should buy his software.

Why don't you go there and discuss it with him?

 

[mansr]

Why don't you go there and discuss it with him?

I was banned in the Great Purge.

 

[q3cpma]

If you've had hands-on or listener experience with modified/optimised equipment you will know that simple improvements in the PSU or grounding can have consequential, audible as well as measureable, impact.

Come one, this is getting ridiculous. Where's the proof for such thing? Also, the proof that this can't be measured.

 

[Sgt. Ear Ache]

If you've had hands-on or listener experience with modified/optimised equipment you will know that simple improvements in the PSU or grounding can have consequential, audible as well as measureable, impact.

Oh?

 

[tuga]

I was banned in the Great Purge.

Really? Is it veering to the "other" extreme?

 

[tuga]

Come one, this is getting ridiculous. Where's the proof for such thing? Also, the proof that this can't be measured.

I think that you've missed the word measureable...

 

[Sgt. Ear Ache]

I think that you've missed the word measureable...

The really important word in all this is "audible."

 

[q3cpma]

I think that you've missed the word measureable...

And you the word proof. Anyway, I don't really care about solved problems like these: balanced connections and SMPS have existed for a long time now.