Master Complaint Thread About Headphone Measurements

[Heinrich]

I smooth my measurements less than research/target to see more detail. Going beyond that will give you perceptually false information because our hearing bandwidth enlarges as frequencies go up so a lot of variations are not audible. This is true for both speakers and headphones.

Sure, but for headphones the elder Harman reference is smoothed. So you never know if deviations are to be appreciated or should be considered to be flaws. The new Harman, not officially released yet as far as I know, shows more features. It allows to match the detailed headphone's output better.

You are talking about a research project, not a review.

That's very much true. Wasn't it done already, I wonder.

 

[solderdude]

Headphone measurements vary wildly—blame rigs, couplers, positioning, and personal ears!

That's why one should use one standard testing method/fixture/target and not change that so comparisons to that standard are valid.
This is what @amirm does.
True that positioning, pad condition, clamping force, and leakage can yield variations in measurements even on the same fixture.

The fact that other measurebators may use different standards, fixtures, targets etc makes it that you cannot compare measurements that easily.

And then there is the correlation with what is heard by an owner which may well differ from what is measured because of:

• product tolerance/variation
• positioning on the head
• seal (glasses, hair, way of wearing, head-size/geometry)
• listening level (non-linearities in hearing which is SPL dependent)
• possible issues with high output resistance (tube gear, (AV)Receivers, integrated amps and even headphone outs from external sound cards/amplifiers)
• pad wear/clamping force
• personal taste
• recording quality/tonality

That's why there can be (but doesn't have to be) a discrepancy between what's heard and measured...

 

[oleg87]

I think the subjective part like imaging and sound stage is also something necessary because there is still no way to measure those properly.
A place to vent or discuss issues with headphone measurements like inconsistencies, rig differences, or subjective vs. objective debates. Share your thoughts or compare notes!

IMO they're not just unmeasurable but also have basically no value as points of discussion. When someone says "soundstage" I have no idea if what they mean by it is what I mean by it.

 

[solderdude]

Consistent standards and transparency in headphone measurements are key to fair comparisons

There is also placement on the fixture, pad wear/aging, seal, product variance, reporting, tests performed, targets, smoothing.
Then there is the choice of test fixture (all having the same options) and choice of targets/standards.
Which fixture and why ? and should all testers get git of their fixtures and buy that fixture... what happens if a new fixture comes out ?
Which levels for testing and why.
Which specific tests at which level ?
Which amplifiers are used.

Are listening tests still important ? What recordings ? How does one know if their hearing is representative ?

So... just take headphone measurements for what they are... indicative for tonality between roughly 100Hz and 6kHz at certain measured level(s) and only somewhat comparative when done by the same tester on the same fixture and hoping operation of it all is done properly and consistently.

 

[scorlee]

Is it possible to evaluate some slightly cheaper devices?

 

[staticV3]

Is it possible to evaluate some slightly cheaper devices?

Sure, Amir doesn't care how much the headphones cost.

He just tests whatever ASR members and manufacturers send in.

 

[FireLion]

Would there be a way to test that excessive EQ is damaging drivers? If what is mentioned in the video linked here is true and with EQ on the rise, people would be experiencing higher rates of failure if done incorrectly.

I prefer not to EQ, I like to just put the headphones on and press play. However if the benefit is substantial then its worth adjusting. The person in the video recommends not going more than 3db on any given frequency, sort of a less is more approach.

 

[solderdude]

In most cases it is the bass that needs some boost, especially so for open models.

To get to the same perceived loudness (which is determined by content in the middle frequency range) the driver gets a lot higher voltage at lower frequencies.
The lowest frequencies require the largest excursion.
Music has the biggest signal for lowest frequencies as well anyway.

To get a 6dB bass boost you will need 2x higher voltage (and thus excursion) and requires 4x as much power.
To get a 10dB bass boost you will need 3.2x higher voltage (and thus excursion) and requires 10x more power.
Then the sound will be about equally loud but with more impressive bass which does not add much to perceived loudness (it does a little).

How potentially problematic that is depends on a lot of factors.
Whether a substantial bass boost (at higher listening levels) thus can shorten the life of a driver depends on:
power rating, excursion limits, driver construction etc.
The weak point for planar drivers is usually the point where the contacts are made to the 'traces on the membrane'.
For dynamic drivers it usually is either the power rating (wires heating too much) and/or breakage of the wire from the edge of the driver to the voice-coil.
This means large excursions can increase the chance of driver failure.

This cannot be generalized.
I agree that it is probably best not to have to resort to extreme EQ in order to get a good sound. This is rarely good for fidelity.
Get a headphone that is comfortable, is drive-able with the source(s) you have and sounds good already without any EQ so it only needs some minor corrections.
Most drivers have no issues with small corrections.
When a driver needs a lot of correction it usually has more issues than just some frequency response ones.

 

[amirm]

Would there be a way to test that excessive EQ is damaging drivers?

In the absolute, no. But you can get an idea from the distortion tests. If the relative THD% is already shooting through the roof in bass, boosting that area again will stress the headphone even more. Such stress though would be clearly audible and level dependent. If you are hearing crackling, extra distortion, etc., then I would moderate the filter in the bass region or disable completely. I routinely comment on this if it is indeed an issue.

 

[Chagall]

So, in other words - more classical, less EDM.

A high-pass filter at 30–40 Hz can be a smart move if listening loudly or using bass boost. At those frequencies, driver excursion is high, but there’s very little perceived loudness.

 

[amirm]

So, in other words - more classical, less EDM.

I have not found EDM to be a problem. It has strong upper bass energy but not much, if any, sub-bass. Some orchestral music, especially movie soundtracks, can have a ton of sub-bass so that would worry me much more than EDM.

 

[Chagall]

I have not found EDM to be a problem. It has strong upper bass energy but not much, if any, sub-bass. Some orchestral music, especially movie soundtracks, can have a ton of sub-bass so that would worry me much more than EDM.

Yeah ok! More Bach, less Hans Zimmer

 

[Curupira]

Would there be a way to test that excessive EQ is damaging drivers?

Amirm already stresses the headphones a lot with the bass shelves he introduces and the volume he tests (up to 114 dB). In those scenarios, some headphones like the Focals and electrostatics tend to crackle, indicating the upper limit of the driver. Others just produce a ton of distortion. As amirm stated, you'll see if the EQ will stress that range with the distortion plots.

Some manufacturers already test damage potential when they indicate the maximum current/power figure they should receive. With that figure, through a calculator and ohms law, you can figure out the SPL produced at 1khz and then extrapolate to the rest of the known FR and see whether you're exceeding that loudness for the relative points you're boosting (eg, if you're testing a linear headphone, from bass to pinna, that saturates at 114dB@1khz, then adding in a 10dB bass shelf might be stressing the driver too much if you already listen too loudly).

 

[solderdude]

There is no problem using (heroic) EQ at any other listening level than 'uncomfortable loud'.
With 'uncomfortable loud' I mean people with normal to moderate hearing loss that feel the urge to dial the volume down within 1 minute of listening at that level/
Soooo.... I would not worry about the longevity of a headphone driver when using EQ in practice.

As Amir already stated... if you have a headphone that is tested by Amir and it shows substantial distortion < 200Hz, has considerable bass roll-off and a low power rating then it may be wise to not play very loud while using heaps of EQ in the bass.

 

[olds1959special]

My Apogee Duet 3 is a good match for my HD650's and with this combo I feel don't need EQ.

 

[somebodyelse]

Yeah ok! More Bach, less Hans Zimmer

Depending on the Bach - for example the Toccata & Fugue in D minor has 16Hz pedal notes if the organ has it, and if they make it through the recording process - you won't find much of it on vinyl because of the high pass filter in mastering and in the rumble filter on playback.

 

[thewas]

Sean Olive and Dan Clark are presenting at the AES Show 2025 Long Beach in October 25 a new headphone target curve defined for the B&K 5128 and GRAS45CA-10 test fixtures:

A Preferred Headphone Target Curve Defined for the B&K 5128 and GRAS45CA-10 Test Fixtures
Saturday October 25, 2025 3:30pm - 4:00pm PDT 103B

The Harman around-ear (AE) and on-ear (OE) headphone target specifies a preferred frequency response using a GRAS 45CA test fixture with the older RA045 ear simulator and a custom pinna designed by Welti (2015) to better match human headphone leakage effects. This target curve is often misapplied to measurements made on other test fixtures with different acoustic transfer impedances, leading to errors and misinterpretations. We measured 27 headphones on the original test fixture and two newer ones, the B&K 5128 and GRAS 45 CA-10. From these measurements, a correct target curve can be calculated to account for fixture differences, enabling more accurate assessments. To validate the method listening tests were conducted using two headphones with low and high acoustic impedances, measured and equalized on three fixtures using the same fixture-specific target to evaluate sound similarity.

AES Show 2025 Long Beach: A Preferred Headphone Target Curve Defin...

View more about this event at AES Show 2025 Long Beach

aesshow2025lb.sched.com

 

[CedarX]

a correct target curve can be calculated to account for fixture differences

Strictly speaking, that would still be an approximate, no? There is no guarantee that the correlation between two fixtures is a simple “function” (in the mathematical sense). Though, if the differences between the fixtures are not to big, this approach certainly works.

 

[Curupira]

Strictly speaking, that would still be an approximate, no? There is no guarantee that the correlation between two fixtures is a simple “function” (in the mathematical sense). Though, if the differences between the fixtures are not to big, this approach certainly works.

Yes, it's a mathematical shortcut validated by some empirical testing. To precisely get the OE equivalent on the 5128, they would have to redo the entire statistical testing.

Given this and the fact that the curve is heavily smoothed, I think "close enough" suffices

 

[Robbo99999]

Sean Olive and Dan Clark are presenting at the AES Show 2025 Long Beach in October 25 a new headphone target curve defined for the B&K 5128 and GRAS45CA-10 test fixtures:

A Preferred Headphone Target Curve Defined for the B&K 5128 and GRAS45CA-10 Test Fixtures
Saturday October 25, 2025 3:30pm - 4:00pm PDT 103B

The Harman around-ear (AE) and on-ear (OE) headphone target specifies a preferred frequency response using a GRAS 45CA test fixture with the older RA045 ear simulator and a custom pinna designed by Welti (2015) to better match human headphone leakage effects. This target curve is often misapplied to measurements made on other test fixtures with different acoustic transfer impedances, leading to errors and misinterpretations. We measured 27 headphones on the original test fixture and two newer ones, the B&K 5128 and GRAS 45 CA-10. From these measurements, a correct target curve can be calculated to account for fixture differences, enabling more accurate assessments. To validate the method listening tests were conducted using two headphones with low and high acoustic impedances, measured and equalized on three fixtures using the same fixture-specific target to evaluate sound similarity.

AES Show 2025 Long Beach: A Preferred Headphone Target Curve Defin...

View more about this event at AES Show 2025 Long Beach

aesshow2025lb.sched.com

That's fantastic, thanks for the update! I'd be curious to try EQ'ing my headphone to whatever new target (I imagine a small adjustment) they come up with for the GRAS! I'm also pretty damn curious to see exactly what they do come up with in terms of differences and indeed B&K too. I may well try to graph the difference between old Harman GRAS Target and new Harman GRAS Target so that people can just apply a small set of filters to their existing Harman EQ's to convert it. Do you know where it's gonna be publicised, is it gonna be streamed on YouTube or anywhere?