Flat response, headphones, and Godel's theorem

[bentoronto]

The sound reaches each ear after quite a variety of acoustic influences. The factors that lead to a judgment of loudness across the freq compass are variously influenced by these factors and not in any well-understood way, even if many people would agree on the loudness of frequencies, either as test signals or with music playing, in the same room.

Test tones and mic traces meant to capture the stimuli at the ears can't do much justice in assessing the human response in this complex situation. Makes me think of Godel's Theorem. Therefore, there is general agreement that it is just accidental when some concept of mic measurement results in system settings that seem right on music.

Is there the same kind of dispute about headphone flatness, say with an Etymotic in-the-ear kind of headphone? In this case, the stimuli do not have much influence like the way music in a room has.

What I'm saying is, we are far from having a system of measurement which can do a good job in setting relative loudness across the freq band. (And that's without considering the Fletcher-Munson issues.)

[Robh3606]

Therefore, there is general agreement that it is just accidental when some concept of mic measurement results in system settings that seem right on music.

Hello Ben

General agreement? Among who?? I would quess that many here who have used measurements to purposely set-up their systems would disagree. They would argue that without the measurements to help point them in the right direction it would have taken much longer through trial and error.

What I'm saying is, we are far from having a system of measurement which can do a good job in setting relative loudness across the freq band

I disagree with that. I can set-up my system very quickly using measurements and get good sound with repeatable results using different speaker systems in the same room.

setting relative loudness across the freq band

I assume you mean octave to octave balance and and overall curve being flat, tilted or otherwise?

Rob

[bentoronto]

Measurements are very helpful for setting up a system. But my impression is that the consensus at this forum is that tweaking is routinely required after.

[dantheman]

The biggest issue I see with headphones are the lack of recordings available to make a better use of their natural advantages. :O We need more dummies. Dummy heads that is.

Dan

[Robh3606]

Hello Ben

But my impression is that the consensus at this forum is that tweaking is routinely required after.

Tweaking can mean many things in this hobby. If you are saying adjusting speaker toe-in, final placement maybe last inch or two, some room adjustments like a rug as an example I would agree for sure.

Rob

[bentoronto]

Originally Posted by dantheman

The biggest issue I see with headphones are the lack of recordings available to make a better use of their natural advantages. :O We need more dummies. Dummy heads that is.

Dan

Do people have the same issue with headphones? Do they want to tweak an Etymotic or do they accept that as the very definition of flat?

[asbjbo]

Headphones need to build in an approximation of the head and outer ear transfer function to "sound flat". This includes a hefty boost around 2700 Hz. If you measure the frequency response, good headphones are far from flat. The brain calibrates its perception of "flat" to the frequency response received through the attached ears. Now, my ears may be differently shaped from your ears, so the one-size-fits-all-ears approximation in a pair of headphones may "sound flat" to you, but not to me, or the other way around.

BTW: I love my Etymotic ER4's "tweaked" with custom earplugs in molded silicon.


Etymotic Research, Inc. - ER-4 MicroPro Series

[bentoronto]

Originally Posted by asbjbo

Headphones need to build in an approximation of the head and outer ear transfer function to "sound flat". This includes a hefty boost around 2700 Hz. If you measure the frequency response, good headphones are far from flat. The brain calibrates its perception of "flat" to the frequency response received through the attached ears. Now, my ears may be differently shaped from your ears, so the one-size-fits-all-ears approximation in a pair of headphones may "sound flat" to you, but not to me, or the other way around.

BTW: I love my Etymotic ER4's "tweaked" with custom earplugs in molded silicon.

Let me restate some of that in what I believe to be more correct (and perhaps more helpful) terminology and offer something (which I now finally realize) is a premise to this discussion.

It is immaterial how the headphones deliver sound you perceive to be flat. They use an EQ to correct for ear and head considerations unique to that delivery system which works for most people or a custom correction for special people.

The brain isn't correcting for anything - the headphones are. In the ear drivers have a different EQ than circumaural ones and different from loudspeakers. I suppose you could say the loudspeakers are correcting for the room and for themselves. But it is also true they are correcting for being a front-stereo delivery system. So a HT 5.1 system would need different correction.

Because heads differ a bit, the Etymotic EQ and/or ear mold has to differ a bit to deliver sound perceived to be flat, at least for cases where the head is atypical. And roughly the same thing is true of rooms; you need to correct loudspeakers differently for each room since none are typical.

Premise: the goal is to deliver sound perceived to be flat. I don't think there can be any other point of view or at least, none that leads to any coherent analysis. The view that "it is a matter of personal taste" might be considered like soup: you can say you like your soup salty and I can say I like it unsalty. But we can agree, roughly, on how the soup in question compares to a "standard soup" in a comparison bowl. Sometimes you get served soup that is too salty and sometimes you get served recordings that are too bassy.

[Pano]

OK. Still not sure what you're driving at.

[bentoronto]

Originally Posted by Pano

OK. Still not sure what you're driving at.

Thanks for taking posts seriously enough to ask.

I am wrestling - thinking out loud with collective wisdom here - about why there is substantial agreement that when you've done your darndest with a mic, you still need to tweak the EQ to have music sound right. That is very puzzling, at least to me.

From that observation, and given the premise that there does exist an objective criterion called "flat," it is fair to conclude that the essential problem is that there is no system of measurement which seems to characterize the heard-experience satisfactorily for enough people and places. I am optimistic.*

Toole feels he has a way of characterizing speakers well enough to predict their room behavior pretty well. But it is a vast set of anechoic measurements.

*Don't forget, we are talking only about getting the freq compass flat. Which ought to be simple. Getting a front-stereo loudspeaker set to reproduce other parameters of Carnegie Hall is all but hopeless.