• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Master Complaint Thread About Headphone Measurements

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
The only difference is the target trace in the raw FR plot and of course the 'post correction' plot.
The other tests and test results do not change nor is it invalid in any other way.

When you don't like the Harman bass boost simply draw the target curve flat between 20Hz and 200Hz for yourself and see how much the measured response deviates from that.
 

Jabinho

Member
Joined
Jul 11, 2021
Messages
46
Likes
19
Margin of error in this type of research is typically +-10% (with 250 people poll).
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,532
Location
Seattle Area
Margin of error in this type of research is typically +-10% (with 250 people poll).
Where the heck do you continue to get this stuff? This is a preference test where to get confidence, you have to perform ANOVA analysis. Here are the results for the overall tonality:

1632207089490.png
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,532
Location
Seattle Area
So not a lot difference.
You call this "not a lot of difference?"
1632207258546.png


You are posting some of the most uninformed opinions in this thread. I suggest stepping back, learning more about the topic and then commenting.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Margin of error in this type of research is typically +-10% (with 250 people poll).

+/- 10% error = +/- 1dB. ;)

The bass lift in Harman target changed more than that over the years.
 

TK750

Active Member
Joined
Apr 19, 2021
Messages
228
Likes
408
Location
UK
Genuine question, could one argue having an 89%/11% male/female split is a bit problematic? Would it not be better to have it closer to 50/50?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
89/11 seems like a better ratio as on this and many other audio related fora so representative ?
 

TK750

Active Member
Joined
Apr 19, 2021
Messages
228
Likes
408
Location
UK
Hmm I feel like a preference curve should account for all, not just male dominated forums. Whilst most don't currently, one could hope that in future mainstream manufacturers making earbuds/headphones for the masses (and therefore a much more even gender split) would use or at least consider the harman research/preference curve in their tuning.
 

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
Fwiw, Dr. Olive explains a little more about how and why there were more males in the studies in this recent video presentation...


I think it basically boils down to the fact that there are more male engineers, salespeople, etc. in the trade and in acoustical engineering than there are females. Since alot of the testing was done with Harman employees (and engineering students?). He also breaks down some of the preferences along demographic lines. My recollection is that females and older males preferred less bass than the generic Harman curve, and younger males preferred more bass.

There are apparently some other studies that also link a preference for more bass to more aggressive personalities, which Dr. Olive also briefly refers to as well. That could be just of a function of age, gender, and better hearing in younger versus older males though. So I wouldn't necessarily read too much into it.

This may seem a bit counter-intuitive, but younger males may seem to prefer both more bass and treble, because they have less hearing loss in the upper midrange frequencies. And prefer that area to be more withdrawn than older males. If that is the case, then the natural tendency might be for them to raise both the bass and treble to some extent to make the upper mids less prominent by comparison. Older males are probably still going to want even more treble (and also mids) than the younger males though, because they have more progressive hearing loss in the higher frequencies.

Younger males may also prefer to listen at lower levels than some older males, and need more Fletcher-Munson (ie boosted bass and treble vs. the mids) as a result to achieve a more perceptually linear sound. The studies did not really take these differences in loudness into account though. Because most were done at close to a reference volume in the 80 to 82 dB range (according to Dr. Olive). This could also be influencing their decisions to purchase headphones with less prominent mids and a bit more prominent bass and to some extent treble though.

The studies actually confirm that not everyone will prefer the generic Harman target. I have my own theories as to why this may be, some of which I've also shared in the above topic. Imo, it's probably mostly related to differences in hearing loss and the circle of confusion / improper mastering of audio content.

Some of it could also be tied to differences in anatomy. Especially the shape and size of the head and body, which will effect the tonality of the sound to some extent before it reaches your ear when listening to loudspeakers, but not with headphones. That's probably something that also needs to be looked at a bit closer.

More studies need to be done though to better understand and get to the real roots of all of this imho.
 
Last edited:

adude995

Member
Joined
May 17, 2019
Messages
15
Likes
5
Location
aut + fin
I'm the guy that made anatomically accurate ears a good 3 years before the B&K 5128 came around... On my own.
You're the guy that put a capsule on a piece of plywood.. and now gets offended that somebody has the audacity to suggest there are better ways to do it..
In case your "accurate" ears are not perfect, without any measurement about its frequency and resonance behavoir and not knowing what the curve should look like, your ear construction is worthless.
Any microphone is plywood measurement is more meaningful.
 

adude995

Member
Joined
May 17, 2019
Messages
15
Likes
5
Location
aut + fin
The one complaint I have to make about procedere of the measurement:

Showing the group delay from a measurement with a pinna is not too accurate. The ear itself has many resonances which cause additional group delay.
The answer would be an additional measurement without an ear in it's signal flow. I know, that would mean extra work and I'm not asking you to do that.
But also phase measurements would be able on such kind of fixtures.


Now to my suggestion:

Many people do use EQ, there is even an conclusion about the sound of the headphone when EQed in the reviews.
Measuring the Amplitude chart after EQing does not make sense. If it is not flat, you have done something wrong.
What we can measure very well are the resonances after have applied the EQ.

Since resonanceS are crucial concerning the sound, we need to look at them precicely.
And I by resonance I mean missing attenuation which can not be fixed with an EQ.

Again these measurement must be made without an ear since it causes resonances itself.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,532
Location
Seattle Area
I am not following what you are asking. To be clear, there is no option to measure anything without the "ear." The artificial pinna is very delicate and can be damaged if taken on and off. And it is not cheap. Creating an entirely different fixture is not in the cards either.
 

adude995

Member
Joined
May 17, 2019
Messages
15
Likes
5
Location
aut + fin
Actually the pinna can be taken off.
Just don't pull on the pinna itself.

Again, I don't want to sound impolite, just talk about what could be possible.
And it is possible to exchange the pinna on the 45ca with an 1" microphone.
Measurements done with that mic don't have resonances we want when measuring the amplitude but do not want when measuring the phase (resonances and group delay)

In theory, it would be possible to have the pinna on one side and the microphone on the other side.
Of course that would mean, that both sides could not be measured simultaneously anymore.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,532
Location
Seattle Area
Actually the pinna can be taken off.
I didn't say it wasn't. I have already taken it off twice to calibrate the mics. In one occasion, it rotated the mic and caused its connecting delicate coax wire to come out! Had to rip it all out to repair and put back together.

You haven't made a case to measure anything of value yet so even if this were easier to do, I would not be signing up for it.
 
D

Deleted member 16543

Guest
In case your "accurate" ears are not perfect, without any measurement about its frequency and resonance behavoir and not knowing what the curve should look like, your ear construction is worthless.
Any microphone is plywood measurement is more meaningful.

Based on this and your following comments all I can say is that you obviously have some reading to do in acoustics and how geometric boundaries are an intrinsic part of the final measurement result.
The geometric boundaries act like a distributed parameter, not a lumped one.
This is why measuring with or without the pinna doesn't work like regular mathematics addition.
Either you measure with the pinna (and canal) to get an accurate representation of the pressure wave hitting the listener's eardrum, or you measure something psychoacoustically quite meaningless like the pressure hitting a flat wooden surface.

I also suggest you look into measurement tools that account for and correct phase response, and provide a measured curve more in line with how humans can perceive different reflections arriving at different times (but of the same frequency) to the eardrum and lump them together into a single perceived sound intensity.
 

adude995

Member
Joined
May 17, 2019
Messages
15
Likes
5
Location
aut + fin
First of all, want to make clear that I understand why we measure the amplitude of a headphone on an ear-canal.

What we must not forget, is the fact that that hump around 4-5kHz is not a property of the headphone itself but is an effect of measuring at the earcanal.
The headphone should be more or less flat in that area. This is also what Sean Olive said in a interview.
(Questions starts at 1:40, quotation at 2:30)

But the earcanal not only amplifies certain frequencies (due to resonances or reflections), it also changes the phase, again due to resonances an reflections.

While its nice to look at an graph with an actual and ideal frequency response, there is no 'ideal impulse response' on an ear canal fixture as well as an ideal group delay /phase response.
The pinna's and the ear canal's phase transfer function is just overruled by the one of the headphone itself, but still, the headphone's phase is of our interest.
(I know headphone itself is critical to say)

To make visible what I am trying to say, I made two measurement of the same headphone (k702)
One of them is an a flat plate and one of them on a self made ear-fixture which is not accurate in any way since I don't have a targent curve.

Magnitude of botch measurements, only to get an impression:

amplitude k702 comp.png


Impulse response of the measurement of k702 on ear-fixture:

impulse k702 silcear.png


Impulse response of the measurement of k702 on flat plate:
impulse k702 flat plate.png

Phase measurement of k702 on ear-fixture:
phase k702 silcear.png

Phase measurement of k702 on flat plate:
phase k702 flat plate.png


As seen in the pictures, the impulse response and the phase, which are both in our interest, are manipulated by the ear-fixture in a big way.
That's why I came to the conclusio, that for those measurements a different fixture, or at least a different sound pressure sensor does make sense.

I wanted to post this, not because I want to tell anybody that he is false and I am right, I just really like the scientific approach of this community has and thought we can have an interesting discussion without making suggestions to each other about any readings we have to do.
Actually I do some readings right now for the research at my university.

I really admire your work Amir, if I made a wrong impression on that.
 
Last edited:
D

Deleted member 16543

Guest
First of all, want to make clear that I understand why we measure the amplitude of a headphone on an ear-canal.

What we must not forget, is the fact that that hump around 4-5kHz is not a property of the headphone itself but is an effect of measuring at the earcanal.
The headphone should be more or less flat in that area. This is also what Sean Olive said in a interview.
(Questions starts at 1:40, quotation at 2:30)

But the earcanal not only amplifies certain frequencies (due to resonances or reflections), it also changes the phase, again due to resonances an reflections.

While its nice to look at an graph with an actual and ideal frequency response, there is no 'ideal impulse response' on an ear canal fixture as well as an ideal group delay /phase response.
The pinna's and the ear canal's phase transfer function is just overrules the one of the headphone itself, but still, th eheadphone's phase is of our interest.
(I know headphone itself is critical to say)

To make visible what I am trying to say, I made two measurement of the same headphone (k702)
One of them is an a flat plate and one of them on a self made ear-fixture which is not accurate in any way since I don't have a targent curve.

Magnitude of botch measurements, only to get an impression:

View attachment 155736

Impulse response of the measurement of k702 on ear-fixture:

View attachment 155737

Impulse response of the measurement of k702 on flat plate:
View attachment 155738
Phase measurement of k702 on ear-fixture:
View attachment 155739
Phase measurement of k702 on flat plate:
View attachment 155740

As seen in the pictures, the impulse response and the phase, which are both in our interest, are manipulated by the ear-fixture in a big way.
That's why I came to the conclusio, that for those measurements a different fixture, or at least a different sound pressure sensor does make sense.

I wanted to post this, not because I want to tell anybody that he is false and I am right, I just really like the scientific approach of this community has and thought we can have an interesting discussion without making suggestions to each other about any readings we have to do.
Actually I do some readings right now for the research at my university.

I really admire your work Amir, if I made a wrong impression on that.

Why would you say that your measurement is not correct 'since you have no target curve'?
If you made no mistakes in the procedure to measure the headphones, the measurement IS correct.

It might not be useful to its full potential, since you don't know how far it is distant from a target curve, but that's a whole another story.
I have written down in the past a couple posts on what the procedure to find this target curve should be in my opinion, for rigs with ear canals.
However, scientific as this forum might be, it's far from being an acoustic research type of forum, so we have no target curve for a rig with ear canals. Not even a ballpark one, which would still be accurate only to a point in your case, since every rig should have its own target curve.

Anyway, I applaud your inclusion of the phase response. It's about time people start giving more importance to this.
Phase response is audible, and it needs to be measured.
 

adude995

Member
Joined
May 17, 2019
Messages
15
Likes
5
Location
aut + fin
Why would you say that your measurement is not correct 'since you have no target curve'?
If you made no mistakes in the procedure to measure the headphones, the measurement IS correct.
I said that, because if I now make a measurement on my ear fixture, it will definitly look different than on another fixture.
And since we don't want to measure different silicon ears, but the headphones, we need to take the ear part out of the measurement, which i don't know and is not easy to identify in generel.
which would still be accurate only to a point in your case, since every rig should have its own target curve.

Anyway, I applaud your inclusion of the phase response. It's about time people start giving more importance to this.
Phase response is audible, and it needs to be measured.
Exactly, and because the impulse response is a compound of an amplitude and it's phase (or the other way round) either the impulse response or the phase would be nice to know.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
every rig should have its own target curve

Yes, and even for different types of phones.
For IEM's and over-ear headphones that should differ as well and maybe even on-ear will be slightly different from over ear.

I'd like to differentiate between target curve and correction myself.

To me the correction curve is what is needed to 'undo' the changes that have been made by the measurement rig acoustic properties to arrive at a 'compensated measuring flat' curve (like Crinacle seems to do).
A target curve (IMO) is the correction curve + tonal correction (think added bass in the Harman curve or any room curves folks like in there) or the one Optimum hifi curve.
 
Top Bottom