• 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!

ASR Getting Into Measuring Headphones!

Omid

Member
Joined
Nov 8, 2019
Messages
22
Likes
14
@amirm I wonder if measuring a few loudspeakers at ~10 feet with this system might provide some interesting data (sort of a summation of direct and indirect sound). I realize there are no standards to compare to, but measuring a few well behaved speakers and some crappy speakers might yield interesting results... Just a thought.
 

YSC

Major Contributor
Joined
Dec 31, 2019
Messages
3,194
Likes
2,570
@amirm I wonder if measuring a few loudspeakers at ~10 feet with this system might provide some interesting data (sort of a summation of direct and indirect sound). I realize there are no standards to compare to, but measuring a few well behaved speakers and some crappy speakers might yield interesting results... Just a thought.
this really looks interesting! though with this system without a proper head simulation shape (for the head transfer function?) it might varies a bit with actual person. but then it should provide us with a sense of how a great measuring speaker in the listening position with the room measured FR and then compare that to the headphone measurement system, so we will know how well that fit into the Harman curve or not?
 

wemist01

Active Member
Joined
Jun 18, 2019
Messages
110
Likes
86
Location
Chicagoland
Hopefully you will be in a position to measure the results of a few of the EQ products- you could even make that a standard feature of your reviews. I use Sonarworks True-Fi (not their most current product, but it has what I need), which really helped my AKG-K7XX's, Foxtex TH-X00s and Audeze LCD-XCs. All of these needed significant help with high frequencies, with the Audeze sounding the weirdest without EQ and the best with. Looking forward to the reviews!
 

MKreroo

Active Member
Joined
Feb 21, 2020
Messages
143
Likes
67
Nice to hear this, I supposed it will be great to perhaps also take Oratory1990's data in as a comparison?
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,444
Location
Seattle Area
Nice to hear this, I supposed it will be great to perhaps also take Oratory1990's data in as a comparison?
Yes, when I measure I seek out others to see how they correlate.
 

BolusOfDoom

Member
Joined
Jan 19, 2020
Messages
54
Likes
139
Would it be possible to measure headphone frequency response compared to headphone amplifier power to figure out if and when bass is lacking because of insufficient power?

I ask because some of the headphone amplifier reviews here have noted a subjective lack of bass attributed to insufficient power.
 

AudioJester

Addicted to Fun and Learning
Forum Donor
Joined
Jan 7, 2020
Messages
913
Likes
1,213
Wow, awesome Amir!
Made me remember I have not donated in a while - well fixed that.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
we are going to be officially measuring, reviewing and characterizing earphones and headphones


Nobody has mentioned soundstage, image, or imaging in this thread, but it often occurs when someone asks about the user experience, as one of the "characteristics".

So...

I wonder what, if anything, can be gleaned from the measurements on that front.
 

bobbooo

Major Contributor
Joined
Aug 30, 2019
Messages
1,479
Likes
2,079
I have looked at it, it allows >100 for the score (Oratory calculates it and for some EQ’d models they get like 106; don’t know why the loudspeakers use the 0-10 scale and headphone use 0-100.

114.49− (12.62*SD)−(15.52*AS)
SD: standard deviation from 50Hz-10kHz
AS: abs of slope of linear regression line from 50Hz-10kHz

Besides log spacing, it doesn’t even treat bass different, whereas the speaker formula does. With just 2 variables, I’m surprised by the stated accuracy, oddly same as speakers, 0.86.
130 people with 31 headphones.

Yeah the max possible score for over-ear headphones is 114.49 as per the formula. For the in-ear headphone formula however, it's normalized to max 100 (or near enough). There's also an additional variable (AE, average error), and different frequency ranges for the variables:

100.08 - (8.5*SD) - (6.8*AS) - (3.48*AE)
SD: standard deviation from 20Hz-10kHz
AS: abs of slope of linear regression line from 20Hz-10kHz
AE: average error of abs deviation of error curve from 40Hz-10kHz

Slightly higher correlation of 0.91 for the in-ear formula, with 71 listeners and 30 headphones. As I said in this post, full spreadsheets for both the over-ear and in-ear formula calculations have been made freely available by Listen Inc. courtesy of Harman under the 'Get Sequence' section here (a short registration form just needs to be filled out to download them): Prediction of Listener Preference of In-Ear Headphones using the Harman Model – Listen, Inc. So just data entry is required really, no extra work this time :)

I will repeat from my following post though, the above Listen Inc. page says under Directions > Notes on Running the Sequence:
  • The IE target curve was developed using a G.R.A.S. RA0045 IEC60318-4 compliant coupler. Other common fixtures for testing IE headphones such as B&K 4128, B&K 5218, G.R.A.S. KEMAR, etc. have not been validated against the target and may not produce results that correlate 100% to the predictive model.
  • The AE/OE target curve was developed using a G.R.A.S. 45CA headphone test fixture with a custom pinna. Other common fixtures for testing AE/OE headphones such as B&K 4128, B&K 5218, G.R.A.S. KEMAR, etc. have not been validated against the target and may not produce results that correlate 100% to the predictive model.
So it looks like in order to have the most accurate preference ratings, the RA0045 coupler should be used (as Oratory does), not the newer RA0402 'hi-res' coupler. The latter could result in not only less accurate results for preference ratings, but less accurate EQing to the Harman target too, and the difference is not insignificant, with potentially audible-level differences when EQing in the treble from around 6kHz up, with a 6dB difference at 10kHz:

index.php


Using the RA0402 hi-res coupler for measurements could then result in falsely EQing the treble up to 6dB too hot when aiming for the Harman target (in addition to lowering the 0.86 / 0.91 correlation of the calculated preference ratings using such measurements).
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,580
Likes
38,282
Location
Gold Coast, Queensland, Australia
Great news @amirm Looking forward to you uncovering some hidden gems and perhaps raining on a few parades as well.

I would like you to consider suitable levels when doing the subjective part of the headphone reviews, so as to not cause yourself permanent hearing damage. When I hear comments like "my ear lobes were vibrating" it worries me...
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
So it looks like in order to have the most accurate preference ratings, the RA0045 coupler should be used

Doesn't this sound a bit like the tail wagging the dog?

As noted on the G.R.A.S. website for the RA0045 - It has a step half-wave resonance at 13.5 kHz which effectively limits its usefulness to below at least 10 kHz.

As for speaker evaluation, any computed preference ratings are likely going to be something for a giggle on the side, and not form part of the actual reviews. They are, and no doubt will remain, something of a contentious metric. If the Harman headphone curve was created with a flawed microphone, we simply need to acknowledge that that part of the curve is not fully valid, and not judge any headphone by imposing blind adherence to it. I'm sure Amir is hoping to repeat his efforts with speakers, and build up a publicly available high quality database that exceeds pretty much anything else available. Doing so with the more accurate microphone is most certainly adding to, rather than just repeating, the existing body of knowledge.
If Amir decides to repeat the equalisation efforts that have proven rather successful in speaker evaluations, I'm sure any issues in the computed equalisation curves will become clear pretty quickly. That will also add to the body of knowledge.
Understanding the provenance of the Harman curve OTOH is vital.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,444
Location
Seattle Area
Nobody has mentioned soundstage, image, or imaging in this thread, but it often occurs when someone asks about the user experience, as one of the "characteristics".

So...

I wonder what, if anything, can be gleaned from the measurements on that front.
A lot but I am not prepared to state that right and create an argument about it. :)
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,374
Likes
234,444
Location
Seattle Area
Understanding the provenance of the Harman curve OTOH is vital.
Yeh, there are some issues here I am trying to resolve with Sean Olive. More on this later.
 

Dreyfus

Active Member
Joined
Feb 7, 2019
Messages
247
Likes
296
Location
Germany
The danger with that is I might end up shaving the beard and slicing sub optimal bits of ear off in pursuit of better measurements
Well, it's actually something practically you would do for a better listening experience.

As I mentioned, you may end up with much less bass compared to the average (Harman or any other target) just because of bad sealing properties. Without knowing one could be fooled into believing that he or she is an absolute basshead ...

this really looks interesting! though with this system without a proper head simulation shape (for the head transfer function?) it might varies a bit with actual person. but then it should provide us with a sense of how a great measuring speaker in the listening position with the room measured FR and then compare that to the headphone measurement system, so we will know how well that fit into the Harman curve or not?
The Harman target does already include an HRTF model. They used a dummy head (with their ear simulator inserted) to record the in-room response at the listening position.

Nobody has mentioned soundstage, image, or imaging in this thread, but it often occurs when someone asks about the user experience, as one of the "characteristics".

I'm afraid that we do not have the approriate standards and technical requirements to capture such highly individual phenomenons, yet.


... not the newer RA0402 'hi-res' coupler. The latter could result in not only less accurate results for preference ratings, but less accurate EQing to the Harman target too, and the difference is not insignificant, with potentially audible-level differences when EQing in the treble from around 6kHz up, with a 6dB difference at 10kHz:

Using the RA0402 hi-res coupler for measurements could then result in falsely EQing the treble up to 6dB too hot when aiming for the Harman target (in addition to lowering the 0.86 / 0.91 correlation of the calculated preference ratings using such measurements).
That is on of the many evidences showing that there is not a truth but thousands of varied interpretations of what could / should / is clamed to be the truth.

Also, the term "accuracy" is only considered in relation to a controlled test system with a standardized fixture. As soon as you vary the contact face and the size and shape of the pinna and ear canal (in short: introduce human factors) the whole thing will collapse into a grey area of uncertainty. Measuring frequency responses with a standardized ear simulator can be a completely different thing than the actual listening experience as soon as the anatomy of the individual deviates from the test fixture.

As for the coupler / ear simulator resonances, keep in mind that the data shared by GRAS do only reflect the measurements of IEM, which makes the system a closed tube resonator!
For circumaural headphones you have to calculate the resonances for an open tube with much longer wavelengths (10 mm of coupler tube + ear canal + lateral concha gain). The higher the wavelengths, the lower the frequency of the modes - which will introduce much deeper resonances in practice than those specified by GRAS for the closed resonator case!

The Acoustics of Hearing Aids: Standing Waves, Damping, and Flared Tubes

Unfortunately, there is no satisfying conclusion for such phenomens. The resonances may shift in both level and frequency depending on the length, diameter and shape of the ear canal.

IMO both the old and new 711 / 60318-4 are wrong because they suggest a somewhat completed science that is able to address all the individual factors. The newer one is smoother, indeed. But not necessarily more accurate. But then again ... what is even "accuracy"? ;)

  • The IE target curve was developed using a G.R.A.S. RA0045 IEC60318-4 compliant coupler. Other common fixtures for testing IE headphones such as B&K 4128, B&K 5218, G.R.A.S. KEMAR, etc. have not been validated against the target and may not produce results that correlate 100% to the predictive model.
  • The AE/OE target curve was developed using a G.R.A.S. 45CA headphone test fixture with a custom pinna. Other common fixtures for testing AE/OE headphones such as B&K 4128, B&K 5218, G.R.A.S. KEMAR, etc. have not been validated against the target and may not produce results that correlate 100% to the predictive model.

I am pretty sure that they confirmed and recalibrated their findings after the release of the newer "high-res" simulator. However, with that much smoothing applied I would consider the differences to be rather negligible anyway.

As for the pinna, they mentioned that they adopted the newer anthropometric model right at the time where Welti experimented with the sealing properties of the old and stiff IEC pinna which was actually developed for hearing aids.

Regards
Dreyfus
 
Last edited:

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,471
Location
Land O’ Lakes, FL
Just don't start with Neumann headphones, I don't have the patience to read through that many comment pages lol...
For those unaware:
DA905B96-DDA2-4D96-8D0A-2D9702A5A56B.jpeg


As a comparison:
BCF10748-8891-47D8-BCA7-359C11BC23D4.jpeg


A boost below 200Hz is good, as the Harman curve does too, but that flatness at 2kHz is not.
 

XGEOX

Member
Joined
Apr 14, 2019
Messages
17
Likes
35
Location
Somerset, UK
I am listening to AEON Flow closed headphone post EQ and it sounds so nice

As a user of these I'm looking forward to seeing your EQ preferences. I had mine in the office (back when that was a thing), tuned on the ADI-2. I can't remember the specific adjustments I made but it would be fascinating to see if they come close to your informed/measured EQ.

Hopefully you'll be covering others in my collection in the near future.

As with your speaker reviews, having suggested EQ for headphones would be a great starting point for personal tweaks.

I've increased my Patreon tier to help support this endeavour.
 

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
6,875
Likes
6,673
Location
UK
Well, this is all very encouraging, I'm looking forward to reading the headphone reviews on here & seeing the measurements.....I do believe this has value and especially when reviewed on this site given the more rigorous scientific bent. Yes, as Amir might have been referring to, we don't need long arguments about how relevant headphone measuring is, as I know that is a sometimes hotly debated topic, but I believe everything has already been said on this front in many threads on this site, no need to keep hashing out & certainly not in extended fashion, a quick nod to it once in a while is ok I think, but not pages of discussion. I look forward to the reviews on here.
 
Top Bottom