Master Thread: Are Measurements Everything or Nothing?

[kharan]

No, because it's the critical issue that you are overlooking.


You guys are just making stuff up and stating it as axiomatic. Can we discuss the actual research instead?

In a nutshell, 100% of listeners prefer the Harman Curve to within 1 dB above 300 Hz, but below 300 Hz listeners split into 3 preference groups: 64% prefer the Curve within 2 dB, 21% prefer the Curve modified to -2 dB to -3 dB of bass, and 15% prefer the Curve modified to +3 dB to +6 dB of bass.

The underlined bit IMO is extremely important and gives a clue to how little human preference varies for sound quality in general. And even in the bass it’s the same curve, with tweaks to the level (not to the shape), and 100% of listeners fall into one of the 3 tweaks. IMO overall the message is one of remarkably high consistency in preferences: one only needs to ask a person if he or she is a bit averse to bass, or a bit crazy about bass, or neither, and you can pretty much hand him or her the target curve to look for in headphones. Wow. Even easier, just hand them the standard Harman Curve 100% of the time, and say there is a 1 in 3 chance they might want to apply a bass shelf of plus or minus a few dB to taste. Wow wow. No wonder Dr Toole describes us as stable and consistent measuring instruments.

....and that is despite all our varying head and ear canal variations as individual humans. And the reason is the bit you, Heinrich, want me to "forget about": the brain sets reference for natural sound including one's individual head and ear shape.

cheers

Those bolded wide band level of bass and treble are exactly what I speak of when I mention very coarse band adjustment. As far as I know we don't have testing with fine EQ in the mids and treble.

The variations of ear canal and ear shapes don't matter when we're listening to speakers because the speaker system because you can essentially consider the speakers and the room as creating a sound field that arrives to the ear. In a headphone, the distances are such that the head and ear interact back with the headphones, the head is part of the room so to speak of whereas you can take out the head if you're doing room measurements with speakers.

 

[krabapple]

You guys are just making stuff up and stating it as axiomatic.

For real.

 

[MattHooper]

Enjoying listening to music is where the fun comes.

Well, let’s face it, audiophiles generally have fun with the equipment too. That’s what separates audiophiles from the average person or music lover - is an enthusiasm for audio gear.

That’s what brings pretty much all of us here to shoot the shit about it.

If you’re not having fun why bother?

 

[Heinrich]

About „the brain“ as the target.

No, because it's the critical issue that you are overlooking.

And „no“ in reply. The brain is an apparatus, but the individual is not. As the humanist I am, I would talk of „brain“ if the topic was hearing aids. Stereo, recorded sound in general, are not meant like that. They present an offer for imagination as fodder for the mind. That‘s where our disagreement starts.

You guys are just making stuff up and stating it as axiomatic. Can we discuss the actual research instead?

In a nutshell, 100% of listeners prefer the Harman Curve to within 1 dB above 300 Hz, but below 300 Hz listeners split into 3 preference groups: 64% prefer …

Following the above prerequisite that doesn’t make any sense to me. Taking The Harman research by the word it is about marketing, targeting the audience as a statistical body. That‘s not me. Frankly spoken, to be seen as „brain, automated“ or a part of one of three groups doesn‘t feel right, maybe a bit totalitarian even (no offense intended).

At least with headphones, there are quite obvious reasons to acknowledge personal differences. But, from my perspective, I speak against a white wall for too long now as to hope for a fruitfull conversation. Let alone the very fact that the design of recordings in technical terms rarely optimizes for headphone listening.

Closing the loop, I‘m once and again shocked by fraudulent advertising like the lately given example, which was explicitly addressing ASR. More so it abuses scientific and engineering terms and language in hilarious, maybe disgusting ways. To counter with de-personalization doesn‘t help, me thinks. Could we leave it as that?

Tune it to your liking!

 

[Newman]

Those bolded wide band level of bass and treble are exactly what I speak of when I mention very coarse band adjustment. As far as I know we don't have testing with fine EQ in the mids and treble.

Within 1 dB is not fine enough for you?

The variations of ear canal and ear shapes don't matter when we're listening to speakers because the speaker system because you can essentially consider the speakers and the room as creating a sound field that arrives to the ear. In a headphone, the distances are such that the head and ear interact back with the headphones, the head is part of the room so to speak of whereas you can take out the head if you're doing room measurements with speakers.

Not consistent with the evidence that 100% prefer mids and treble within 1 dB of the Harman Curve for Headphones. Not at all. I suspect you still have not grasped the point about one's own brain calibrating reference for naturalness to one's own physiognomy. If you grasp that, you won't keep posting the above opinion.

There is a level of dysfunction in where you are taking this. I think it relates to the 'Wine Taster' approach to component selection: "Let me try this one. Now let me try that one. Mmmmm." If the headphones are each within 1 dB of the Harman Curve, you are better off choosing on other criteria like comfort etc, because (a) EQ is much more fine-tuneable if you want that last spatula slice of unique, this-is-me personalisation, and (b) recording variations in tonality are going to absolutely swamp the differences you are sipping like wine...by a mile.

 

[Heinrich]

Within 1 dB is not fine enough for you?
Not consistent with the evidence that 100% prefer mids and treble within 1 dB of the Harman Curve for Headphones.

If you please don‘t take any offense, you’re misinterpreting the research of Harman as a company. You focus on what Harman gives you, not on what it should be, if it were independent science. A more critical stance might be in order. It would anyway be a marketing approach to begin with, not reaching out for universal understanding.

As a round-up take an In Ear Monitoring (IEM) device for instance. Those buttons you put halfway into your ear. Changes the ear canal‘s physical properties, which have to be restored—or replicated, by the IEM‘s design. But it is an individual feature, depending on fit to begin with.

Another one, the diffuse field equalization, why? The argumentation in favor of it is weak, to say the least. And again, a personal feature.

Last but not least, the pinna‘s contribution, connected tightly to above mentioned diffuse field versus directional cues, obviously a very personal thing. It makes those peaks ‚n dips in higher treble—IEMs fake it, no pinna involved. Measurement shows these and people get angry. If they are asked to set it to their liking, well, to no avail. Could it be that it is not about liking, but adherence to ‚objective‘ superiority by measured data? Data may and does mislead, if not understood

Harman‘s more or less bass and treble science sells headphones.

Answer the question: should those peaks and dips be there w/ IEMs or not? Then we are talking. ;-)

 

[Newman]

About „the brain“ as the target.

And „no“ in reply. The brain is an apparatus, but the individual is not. As the humanist I am, I would talk of „brain“ if the topic was hearing aids. Stereo, recorded sound in general, are not meant like that. They present an offer for imagination as fodder for the mind. That‘s where our disagreement starts.

If that were true in respect to the issue of audiophile preference, then you, as a unique imagining-mind-not-brain, would have dissatisfaction with every natural sound naturally occurring. More likely you are still not grasping the ubiquity of the way our brains/minds calibrate a reference baseline for natural sound and 'set' it.

Following the above prerequisite that doesn’t make any sense to me. Taking The Harman research by the word it is about marketing, targeting the audience as a statistical body. That‘s not me. Frankly spoken, to be seen as „brain, automated“ or a part of one of three groups doesn‘t feel right, maybe a bit totalitarian even (no offense intended).

At least with headphones, there are quite obvious reasons to acknowledge personal differences. But, from my perspective, I speak against a white wall for too long now as to hope for a fruitfull conversation. Let alone the very fact that the design of recordings in technical terms rarely optimizes for headphone listening.

Closing the loop, I‘m once and again shocked by fraudulent advertising like the lately given example, which was explicitly addressing ASR. More so it abuses scientific and engineering terms and language in hilarious, maybe disgusting ways. To counter with de-personalization doesn‘t help, me thinks. Could we leave it as that?

Tune it to your liking!

If you please don‘t take any offense, you’re misinterpreting the research of Harman as a company. You focus on what Harman gives you, not on what it should be, if it were independent science.

Ohhh, now I see what's going on here. I am interpreting it correctly, but I'm not judging and dissing it negatively enough to suit someone who wishes it weren't true.

A more critical stance might be in order. It would anyway be a marketing approach to begin with, not reaching out for universal understanding.

As a round-up take an In Ear Monitoring (IEM) device for instance. Those buttons you put halfway into your ear. Changes the ear canal‘s physical properties, which have to be restored—or replicated, by the IEM‘s design. But it is an individual feature, depending on fit to begin with.

Another one, the diffuse field equalization, why? The argumentation in favor of it is weak, to say the least. And again, a personal feature.

Last but not least, the pinna‘s contribution, connected tightly to above mentioned diffuse field versus directional cues, obviously a very personal thing. It makes those peaks ‚n dips in higher treble—IEMs fake it, no pinna involved. Measurement shows these and people get angry. If they are asked to set it to their liking, well, to no avail. Could it be that it is not about liking, but adherence to ‚objective‘ superiority by measured data? Data may and does mislead, if not understood

Harman‘s more or less bass and treble science sells headphones.

Answer the question: should those peaks and dips be there w/ IEMs or not? Then we are talking. ;-)

All I am seeing is someone who doesn't like the findings of the best available research, perhaps thinking in his head "that can't be right", and sets out to pontificate on its terribleness, with a bunch of baseless claims that he sees as axioms, backed up with zero evidence.

As I noted in an earlier thread, with a post that Toole tacitly agreed with a 'like', "I implore that we all explore the audio science of sound reproduction with curiosity and integrity, namely: being open to changing our minds when the available evidence contradicts our prior beliefs; being honest with ourselves when we are feeling resistant to new incoming evidence that challenges our presuppositions; disputing evidence with better evidence instead of just arguments; and adopting the current working hypotheses of those researchers in the field with the broadest oversight, at least until contradictory evidence gathers enough weight to change the consensus working hypotheses."

cheers

 

[kharan]

Within 1 dB is not fine enough for you?


Not consistent with the evidence that 100% prefer mids and treble within 1 dB of the Harman Curve for Headphones. Not at all. I suspect you still have not grasped the point about one's own brain calibrating reference for naturalness to one's own physiognomy. If you grasp that, you won't keep posting the above opinion.

There is a level of dysfunction in where you are taking this. I think it relates to the 'Wine Taster' approach to component selection: "Let me try this one. Now let me try that one. Mmmmm." If the headphones are each within 1 dB of the Harman Curve, you are better off choosing on other criteria like comfort etc, because (a) EQ is much more fine-tuneable if you want that last spatula slice of unique, this-is-me personalisation, and (b) recording variations in tonality are going to absolutely swamp the differences you are sipping like wine...by a mile.

I'm talking about the frequency range of the bands used ie the Q factor of the EQ used. The Harman research basically says that people have fairly universal preferences regarding proportionate amounts of bass and treble. That's it.

I perfectly understand that one's brain is calibrating to one's anatomy with regards to what one perceives as natural, thank you. The issue is that headphones, or worse in-ear monitors don't behave the same depending on the head/transducer coupling. Note that I said coupling, it's not just a matter of a different head transfer function. You would think that by applying the same head transfer function to two different pairs of headphones frequency response, you'd get the accurate frequency response at the eardrum, but that is not the case. There isn't one universal head transfer function you can apply to every headphones to get the response at one's ear drums because every unit couples differently with one's head.

 

[ahofer]

If you please don‘t take any offense, you’re misinterpreting the research of Harman as a company. You focus on what Harman gives you, not on what it should be, if it were independent science. A more critical stance might be in order. It would anyway be a marketing approach to begin with, not reaching out for universal understanding.

As a round-up take an In Ear Monitoring (IEM) device for instance. Those buttons you put halfway into your ear. Changes the ear canal‘s physical properties, which have to be restored—or replicated, by the IEM‘s design. But it is an individual feature, depending on fit to begin with.

Another one, the diffuse field equalization, why? The argumentation in favor of it is weak, to say the least. And again, a personal feature.

Last but not least, the pinna‘s contribution, connected tightly to above mentioned diffuse field versus directional cues, obviously a very personal thing. It makes those peaks ‚n dips in higher treble—IEMs fake it, no pinna involved. Measurement shows these and people get angry. If they are asked to set it to their liking, well, to no avail. Could it be that it is not about liking, but adherence to ‚objective‘ superiority by measured data? Data may and does mislead, if not understood

Harman‘s more or less bass and treble science sells headphones.

Answer the question: should those peaks and dips be there w/ IEMs or not? Then we are talking. ;-)

I don’t think impugning the source is very convincing. It’s a shame we don’t have a wider base of it, but that’s where we are.

 

[Heinrich]

I don’t think impugning the source is very convincing. It’s a shame we don’t have a wider base of it, but that’s where we are.

My criticism didn‘t diminish Harman’s contribution. It is only so, their preference screening didn‘t address the basic problems with headphone measurement, as explicated above. Their focus is/was optimizing for the market, which is not a particularly scientific approach.

O/k, they asked for more or less in bass and treble. They could have, but didn‘t, asked for the location, width and amplitude of the 3k hump. Only to show that this part of the model is proper. They could have asked for the relevance of the 10k dip, especially with IEMs and so forth. Bass and treble is tuning to general taste, not solving a problem in understanding. Engineering for the marketplace, not that much science. Fair enough for sure, but has to be acknowledged. That‘s the next time I raise these questions. Don‘t expect too much.

edit: Toole says, that the direct sound of a speaker is most relevant. Then people get measurement equipment at hand and tune the steady state, reverberation all included to a smooth and steady Harman tilt. Bad idea, ain‘t it? Still those guys are happy, because they just believe „the data“. Don’t they know what to „prefer“? Playback is a strange business, indeed.

 

[MaxwellsEq]

Bad idea, ain‘t it?

Instead of endlessly claiming something is a bad idea without stating what makes it a bad idea...

... actually explain why it's a bad idea, using some facts that we can learn from.

 

[pablolie]

My criticism didn‘t diminish Harman’s contribution. It is only so, their preference screening didn‘t address the basic problems with headphone measurement, as explicated above. Their focus is/was optimizing for the market, which is not a particularly scientific approach.

O/k, they asked for more or less in bass and treble. They could have, but didn‘t, asked for the location, width and amplitude of the 3k hump. Only to show that this part of the model is proper. They could have asked for the relevance of the 10k dip, especially with IEMs and so forth. Bass and treble is tuning to general taste, not solving a problem in understanding. Engineering for the marketplace, not that much science. Fair enough for sure, but has to be acknowledged. That‘s the next time I raise these questions. Don‘t expect too much.

edit: Toole says, that the direct sound of a speaker is most relevant. Then people get measurement equipment at hand and tune the steady state, reverberation all included to a smooth and steady Harman tilt. Bad idea, ain‘t it? Still those guys are happy, because they just believe „the data“. Don’t they know what to „prefer“? Playback is a strange business, indeed.

The only thing I have used measurements for was to optimize sub placement. But... I like to look at measurements to check if the equipment caters to my needs.

While I have slightly altered the headphone settings to cater to my personal taste, it was not in order to achieve any "ideal" response, just personal preference - which I don't have to justify to anyone. And like I said, the changes are very marginal.

My concern with more radical EQing that basically overrides the intended behavior/sound/"personality" is that it may push the device into other limitations. Especially because I don't see us measuring the devices as thoroughly after applying said EQ, we just seem to trust the result must be "better". Personally, I prefer to get something that I like a lot to begin with, not something I need to experiment around with much.

But that's a personal choice on my part - and a bit of a rationalization for my preference. I am in no way claiming "science" is on my side, nor that I can *prove* my established personal preference with measurements. I do pick equipment that trends towards linear-ish behavior, listen to it, and if I like it I leave it at that and enjoy the music... I am very happy with my system in my environment, and would only rethink things if there's a major technoology breakthrough that may cater to my way of listening, or when I move again (although I doubt I'd move into a cavernous place that requires something radically different).

 

[Heinrich]

Instead of endlessly claiming something is a bad idea without stating what makes it a bad idea...

... actually explain why it's a bad idea, using some facts that we can learn from.​

Didn‘t think it was needed. Equalizing for a smooth and steady Harman tilt in reverberant sound field will necessarily alter the direct sound field. Because room properties, listener position etc in general don‘t match the speaker‘s directivity properties together with its direct field tuning.

As, by your alias, you are familiar w/ equations:

Ideal frequency response is „1“, ideal speaker is f = 1, but we have

fd = x in direct field, find eq/ e so that

fed = e * x = 1

The room (r) and directivity (d) and equalized direct field (fed) shape the reverberant sound like

fr = fed * d * r

Now you equalize the direct field optimized speaker to Harman tilt in-room w/ equalizer E

frE = E * fr = E * e * x * d * r = H

The equalization to Harman necessarily alters the direct sound. In order to still have an optimal direct field we need

E * d * r = H

Since r is an unknown at least when combined in detail w/ properties of d, the equation rarely holds. Conversely, optimizing for H won‘t automatically adjust e to fit. That will miss the primary target of fed = 1, which according to Toole is the goal, which H is not.

So much on the science of playback gear *g*—maybe I‘m an old rant, but sometimes …

 

[MaxwellsEq]

Didn‘t think it was needed. Equalizing for a smooth and steady Harman tilt in reverberant sound field will necessarily alter the direct sound field. Because room properties, listener position etc in general don‘t match the speaker‘s directivity properties together with its direct field tuning.

As, by your alias, you are familiar w/ equations:

Ideal frequency response is „1“, ideal speaker is f = 1, but we have

fd = x in direct field, find eq/ e so that

fed = e * x = 1

The room (r) and directivity (d) and equalized direct field (fed) shape the reverberant sound like

fr = fed * d * r

Now you equalize the direct field optimized speaker to Harman tilt in-room w/ equalizer E

frE = E * fr = E * e * x * d * r = H

The equalization to Harman necessarily alters the direct sound. In order to still have an optimal direct field we need

E * d * r = H

Since r is an unknown at least when combined in detail w/ properties of d, the equation rarely holds. Conversely, optimizing for H won‘t automatically adjust e to fit. That will miss the primary target of fed = 1, which according to Toole is the goal, which H is not.

So much on the science of playback gear *g*—maybe I‘m an old rant, but sometimes …

That's a bit more helpful thanks.

I'm not convinced that a) the whole Harman curve thing in speakers in rooms is what you think it is, b) people are misapplying it the way you think they are.

My understanding is: in blind testing the majority of people consistently have a subjective preference for loudspeakers which (when measured in an anechoic chamber) have a flat and smooth frequency response and minimal errors such as resonances. Furthermore, smooth off-axis anechoic response also gives a positive nudge to an average group of people's subjective blind tested opinion.

When you put such a loudspeaker in a room, and sit a distance from it, you get a "curve" created by the room and speaker interactions and the fall off with distance.

What I think you are saying is: people have heard about this research and are, without using any critical thinking, using EQ with their speakers in their room to "match" the curves observed by Harman in their research.

But I'm not sure people are doing what you think are doing. Do you have evidence?

 

[Heinrich]

That's a bit more helpful thanks.

I'm not convinced that a) the whole Harman curve thing in speakers in rooms is what you think it is, b) people are misapplying it the way you think they are.

My understanding is: in blind testing the majority of people … Do you have evidence?

I‘m not going to discuss Harman‘s preference science no more. I‘m quite happy with the new standard, which accidentially (lol) confirmed best engineering practice. I‘m really happy, good thing, love it.
Then it is so good to have the Olive score that connects measurement to quality assessment.

The expected „but“ lies in the individual preference (especially headphones, a non-issue with amps, dacs, a delusion with cables), so that people trust their subjective choice more. More often than not all these exceptional perfection isn‘t needed, let alone some alignment to the wider public’s expectations. Because, as I see it, the playback is an invitation to the mind to develop a fantasy, imagination, not illusion for the „brain, automated“. I spent too much time on this all too trivial topic already, @MaxwellsEq. My hobby is deciphering the recording‘s meaning.

 

[rdenney]

That's a bit more helpful thanks.

I'm not convinced that a) the whole Harman curve thing in speakers in rooms is what you think it is, b) people are misapplying it the way you think they are.

My understanding is: in blind testing the majority of people consistently have a subjective preference for loudspeakers which (when measured in an anechoic chamber) have a flat and smooth frequency response and minimal errors such as resonances. Furthermore, smooth off-axis anechoic response also gives a positive nudge to an average group of people's subjective blind tested opinion.

When you put such a loudspeaker in a room, and sit a distance from it, you get a "curve" created by the room and speaker interactions and the fall off with distance.

What I think you are saying is: people have heard about this research and are, without using any critical thinking, using EQ with their speakers in their room to "match" the curves observed by Harman in their research.

But I'm not sure people are doing what you think are doing. Do you have evidence?

Yes, the Harman tilt is an outcome, not a target. Toole said resonances are what we want to avoid, and I could not agree more. Resonances create ringing that I hear and that sound wrong. I’d rather have gaps than resonances.

The tilt happens as a natural consequence of smooth high-frequency falloff off-axis. This happens with live sound, too. If it’s smooth, the reflections will sound like reflections instead of competing sources. As the reflections mix with on-axis sound, some of that falloff will pull the average response curve down smoothly, creating a tilt.

But it’s not a target. If we screw up the on-axis response to account for weird reflections, it won’t sound correct. Axial response and directivity are two different effects that need separate treatments.

Rick “cares less about what others prefer when examining the details” Denney

 

[Purité Audio]

it doesn’t matter what was used in the creation of the record or which equipment was used by the various engineers who worked on it, all we have is the artefact itself.
Keith

 

[J.M. Noble]

You can only measure what you test.

Measuring steady state performance into dummy loads isn't the same as driving reactive electromechanical loads with real-world signals. The former will tell you a lot, of course, but it's not the whole story. I'm not injecting any audiophoolery into the discussion, this is an engineering statement and ought to be self-evident.

 

[ahofer]

When people keep making a lot of claims about things that theoretically may happen, I keep coming back to this (and others).

https://www.worldradiohistory.com/Archive-All-Audio/Archive-HiFI-Stereo/80s/HiFi-Stereo-Review-1987-01.pdf

Page 78

 

[MarkS]

This!

Properly controlled listening tests are completely independent of measurements. If people can indeed hear things that can't be measured, then controlled listening tests should reveal those differences.