• 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!

HBK Headphone Measurement Talks from Head-Fi and Sean Olive

Mad_Economist

Addicted to Fun and Learning
Audio Company
Joined
Nov 29, 2017
Messages
543
Likes
1,618
As I said, design criteria is not proof of accomplishment. Independent testing and research needs to be performed to ascertain that.
Are you suggesting that there are issues with B&K's underpinning research? It seems to agree well with a randomly grabbed paper from an unrelated source, and, of course, underpins the present ITU-T standards for standard human ear Z - I'd expect that the working group at the ITU would have raised an issue if there were an issue...
 

richard12511

Major Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
4,335
Likes
6,700
Bruel & Kjaer (BK), makers of precision acoustic measurement products, had a nice two day virtual conference. I wanted to listen to a bunch of them but had a major conflict. Fortunately, I watched two talks related to headphone measurements:

First one was by Jude Mansilla of Head-fi with the curious title of, "Audio Measurements As Consumers Content."

The second was from Dr. Sean Olive of Harman, going through his standard presentation on headphone preference but also a new research addition: difference between various headphone measurement gear relative to the one they used in their research. Specifically we were treated to the first analysis of BK 5128 which Head-fi has adopted as their standard.

Jude Presentation
I was very surprised and disappointed in most of this presentation. Bulk of time was spent showing and ridiculing the DIY measurement rigs members of his own forum had created. He showed pictures of them and while correctly stating some issues with them, I just could not figure out what he is after. Hobbyist are creating these measurements because headphone companies are not providing them. If Jude wanted to improve things, he should have complained about lack of such measurements from that sector which would be likely to be in this conference, than DIY people. I personally admired the work and creativity people had put in building their various fixtures.

Fortunately the second half of the talk was better in that he showed a bit about how they measure headphones which seems to be following what Tyll did with use of square wave and such. I am personally not a fan of driving headphone or speakers with square wave, especially a low frequency one. This can be hard on the transducer with the long duty cycle essentially being DC. I saw little justification for this method other than a hack to show the frequency response, sort of, using crude FFT.

The other thing he mentioned is that they no longer calibrate at one frequency and instead use white noise. I don't understand the merit of this either as matching the measurements to target needs to be done in a way that relates to the research. Credit to him he asked for feedback from audience but none was provided. Target matching is a visual thing for humans anyway so ultimately it doesn't matter per se.

I finally got an explanation of why they spent so much money on the Herzan isolation chamber. I am able to make measurements with better noise floor than them in my office with no isolation chamber. Answer is that they work in a noisy and busy office building and without noise isolation, they would have fair bit of pollution. In their typical promotion, they always made it sound like this was an essential thing in any measurement system rather than being a requirement in their noisy environment. We are fortunate to live far from civilization so have the quietness most of the time sans the delivery truck and our dogs barking because of that. I do my acoustic testing at night which eliminated this issue. Anyway, having an isolation chamber is not a bad idea but context is important and was not stated until now.

In QA section he was asked what his favorite headphone was. He said the Sennheiser HE-1 ($45,000). He was asked if he had measured it. Shockingly he said no! Gosh that was awkward when you are in a measurement seminar and you don't believe in this stuff to practice it.

Along these lines, he was asked if he felt listening tests were needed in headphones once you have measurements. He said no but could not provide any reasoning why.

Dr. Olive Presentation
As I noted, the meat of the presentation was a new research project to go back and remeasure a 20 headphones against half a dozen HATS and measurement gear. They only disclosed the difference between the modified GRAS 45 that they had used and BK 5128 which happened to be the one I, we are all interested in. The results were fascinating.

The BK 5128 underreported bass frequencies by substantial amount and in a sloping way. The lower the frequency, the less it reported the bass energy. This caused the preference score of a headphone drop 45 points from high 80s down to 40s! There were also some differences in high frequencies but that was not that meaningful or emphasized in the presentation.

They developed a compensation curve for 5128 relative to Harman target but alas, not all headphones showed the same differential. Using this new target, the above headphone showed an error of I think 6 points. But there are others that cannot be fixed this way.

To summarize, there is no way to use the Harman target curve 100% reliably when performing measurements using BK 5128! You can get close for majority of cases if you compensate but not all. And there could be effect on headphones that were not measured in the study.

So the overlaying that Jude does at head-fi with Harman target is totally wrong and inappropriate for bass energy.

As a proof point, Sean actually pulled my measurements of the Dan Clark Stealth and compared it to Jude's:

Mine:
index.php


Judes:

View attachment 156436

He matched the sizes and played with the reference levels to get an apple vs apple comparison. It clearly showed deficiency in bass measurements which he could explain with their research. But Jude's also showed a deficiency in the 3 to 6 kHz which I have circled in read which Sean and team could not explain. The dip does not show up in my measurements. That seems to be a measurement error in Jude's measurement but of unknown cause right now.

This is a fairly big setback for Head-fi headphone measurements based on BK 5128. No benefit was identified in using it, and problems identified with what it does produce. Fortunately some of it can be fixed and I hope Jude adopts the modified curve as to reduce confusion that his measurements can create relative to Harman research and comparison with measurements that the rest of us are performing.

I have been making this point for quite a while that more precision in some measurement doesn't do you any good if you don't have a yardstick to compare it to. And that seems to be the case with BK5128. I think fair bit of the research money in development of that fixture should have gone toward creating realistic target curves for it. Without it, the hype has gotten ahead of its true value putting head-fi in a difficult situation right now.

I felt bad for Jude as Sean's presentation was right after his. He had just finished singing the praises of BK5128 but all he showed for a proof point was the same IEM measurements he has been showing and how an IEM they tested was bright and GRAS measurement didn't show it but BK5128 did (or the reverse, can't recall exactly which).

Sean was also asked about his favorite headphones and he mentioned a couple that were in under $200 range.

When he covered the difference bass preferences, he was asked what they target. He said that the JBL line is boosting the bass a couple of dBs with the assumption that younger crowd were buying so were more inclined to want more bass. AKG line was staying true to the target bass response.

One sour note was Samsung Legal censoring one of his slides. And the fact that the company's policy has become to make sure competitors are NOT taught how to compete with Harman as they disclose their research. This means we will not hear as much as we have heard in the past about their work. I can see why Samsung has taken this position from business point of view, but for the rest of us, it is not a good development.

I hope Samsung management/Legal understand that by sharing their research, their work has become a review criteria which can be very beneficial to them.

That's all I remember to post. I have asked HBK if they will provide on-demand versions of these presentations but I have not heard back.

God, putting Sean's presentation directly after Jude's is kinda awkward. Sean's presentation was excellent with examples and evidence. I too feel bad for Jude, honestly. Wonder if they(the organizers) did that on purpose?
 
Last edited:

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,643
Likes
2,811
Been mentioned before but worth repeating. On iPhone>accessibility>AirPods>audio accessibility settings>Headphone Accommodations and change the sound to how you like,


This is not using parameteric EQ though

It's using built in samples where you select the sample your prefer

That's Better than nothing but a 5 band EQ would be much better

The only exception to this is the Audiogram import feature "Use your audiogram data from the Health app"

I wonder if there's a way to create a custom curve in REW in a format that this iOS feature can use with Airpods
 

JRS

Major Contributor
Joined
Sep 22, 2021
Messages
1,158
Likes
1,003
Location
Albuquerque, NM USA
@MayaTlab @GXAlan Thanks for the response. My question was more towards philosophy and psychology type of question. I was questioning why Harman is doing the best compare to other companies at target curve. Maybe Harman is really gifted or lucky to get it ahead of others, or maybe Harman just gets all the attension because they are the only one showing rigurous papers on how they arrive to the target. I wonder what will happen when they stop providing the proof behind the cruve, or even the curve itself.
I suspect the advantage is real--they have been at work on this for quite a while.

And in the ideal merit-seeking consumer population (where that exceedingly rational,invisible hand notion is at play) the Harman research deserves a strong head start in applying the preference research before others catch on. Maintaining the lead is the problem it seems--you just dropped a bucket of money, enjoyed a temporary advantage and now every company capable of decent msmt is on a level playing field thanks to your magnanimous contribution.

In the past month I have come across at least two very modestly priced speakers which produced superb Klipple results, later confirmed with subjective impressions, and which tracked the Harmon target curve-Polk (not that were ever bad) was one. KlH maybe the other?

In any event, my point is that catch up is easy--at least for the big boys. And were they to mount their own research programs, maybe even pull ahead. But I suspect Harmon/Samsung still has a few tricks up their sleeves to maintain an advantage for a while longer--and that would seem to make secrecy a requirement. Quite obvious, I can see DSP/actives with 5 presets to choose from--no new idea there, and no laborious efforts required.

I for one would spend money with the guys who made the effort to collect the data, provided the prices were close. After all I am acting in my own self-interest. But I would not be surprised if companies with the chops of an SB Acoustics or Aurum Cantus were to produce some giant killers at a seemingly absurd price point. No idea really--this is purely speculation. And to what extent they could succeed in a pricing war with large US/European companies, again no idea, nor would it matter much being a niche market except to us. And it would have to be awfully good to displace the more expensive JBL monitors as a for example.

Less speculative is that however this plays out, we are all big winners (just as was the case with the NRC findings) as a result of this research. My abiding hope, is that the masses, now outfitted with great rigs (there will always be bass heads--or the boom-honk-tinkle devotees as we used to condescendingly refer to those who couldn't grasp the superiority of a Dalquist DQ-10 or Quad ESL), start to create demand for good recordings. I certainly see some of that trend with the headphones stack crowd. That would benefit mankind as a whole, and leave the likes of Drs. Toole and Oliver with a lasting and well deserved legacy.
 

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,643
Likes
2,811
This is not using parameteric EQ though

It's using built in samples where you select the sample your prefer

That's Better than nothing but a 5 band EQ would be much better

The only exception to this is the Audiogram import feature "Use your audiogram data from the Health app"

I wonder if there's a way to create a custom curve in REW in a format that this iOS feature can use with Airpods

I pulled out my Airpods and actually tested this feature for the first time.

As I expected, it is very basic and we have no idea what the EQ adjustments being made are. Better than nothing I guess.

I downloaded Mimi Hearing test on iPhone and did the test and it exported results to Apple Health. Headphones accomodations imported the Audiogram results - now that is very cool and powerful.

The Mimi hearing test is calibrated for a handful of earphones, the Airpods line and a couple Sennheiser models.

If Apple could add 10-band EQ (go look at JBL's mobile app for their bluetooth earphones for the gold standard - EQ is saved inside the earbuds, for ALL apps) - combined with their Audiogram import feature - this would be an incredible combination.
 
  • Like
Reactions: JRS
D

Deleted member 16543

Guest
We don't need "possible explanation." We need real explanation from BK if one exists.

Regardless, if you read the rest of the thread and understood what I wrote as summary in OP, you realize that it doesn't matter. The only research into what we think is a good sounding headphone is from Harman and that research and correlation to measurements do not work for BK 5128. There is an attempt to make some correction but otherwise what I have been saying all along is true: you can't use BK 5128 or any other fixture to draw conclusions with respect to preference for a headphone sound. This is why I bought the GRAS fixture and not 5128.

As to your tired argument of 5128 being more accurate, that is a hypothesis in need of proof. Just because BK has done this and that research doesn't mean any goals have been met. They, the company, needs to do that research and provide the verification. What has gone into design of something is not the same as achieving said results.

Also the whole notion of an average of many people's ears being a model that is most accurate is non sequitur to me. You can't average a bunch of stuff, i.e. filter, and wind up with a specific thing that is accurate. Average temps today may be 60 degrees but that says nothing about the low of 40 and high of 80.

The goal then should be to see how well a measurement predicts listener preference and not how "accurate" it is. We need actionable data. Not random technical goals being met.

Now, it may turn out that a brand new, multi year research with BK 5128 leads to better correlation with its measurements. Until that is done though, the arguments about accuracy are moot. We don't know that it is more accurate, nor better predictor of listener preference.

As far as I am concerned, post measurement Equalization and listening test is key to finalizing the measurement data.
That 'possible' was smore sarcastic than anything else. You don't think that the capsule of the 5128 not having a flat FR would reflect in the measurement of headphones?

As far as the rest goes, I already stated that what you're doing is probably the best possible, practically speaking, considered the readily available target curve. What more do you want me to say? That it is the absolute best possible from here 'til kingdom come? It isn't. Sorry. The B&K is potentially a better rig than the one you use. It is not an hypothesis. It derives from it being able to distort the sound wave less, in respect to what happens in the outer ear for a real listener, compared to the GRAS rig. It's physics, not an opinion.
And while it's true that nobody has an ear canal that matches a canal modeled by averaging many real canals, it is also true that that target curve (when finally produced) will be placed more in the middle of the listener's preference spectrum than the one derived from a rig where half of the canal is a straight cylinder.
That's the value of averaging the canal. It minimizes the error (the deviation from a balanced sound) for everybody, not just the ones who happen to have a canal close to the modeled one.
 
Last edited by a moderator:
  • Like
Reactions: ADU

Resolve

Active Member
Reviewer
Joined
Jan 20, 2021
Messages
212
Likes
531
To me the headphone sounded just horrible, not sure why many claim it to be the "world's best closed headphone". a complete letdown soundwise.
If this is in reference to a video title, I should clarify that it was ended with a question mark, and what was expressed did not definitively conclude that it was, merely that it could be in the same conversation as some of the other examples there. From my perspective, and something that I imagine we'd agree on, is that so much of this particular example comes down to the headphone's positioning. While I didn't find a particular 9khz peak in my most common measurements (in fact every result showed the correct concha dip), I did find some differences in a number of results that were less desirable when the headphones were positioned in different places on the fixture. I even posted about this on a number of forums before doing the video, and I think a few other folks who have heard it found similar things.

What I will say is that the most common result was also the one that I published - in some sense I think you kind of have to do this when conducting multiple measurements of closed-back headphones where the results vary like that, or at least find some indication of what's representative. I think the bigger question there is whether the most common (or let's say 'optimal') result, is the same result that you personally get when wearing it in the most comfortable spot. And this is again why I think so much of the difference found in impressions - or indeed disagreement between subjective impressions and industry standard measurements can be attributed to variation in coupling. I don't think there's ever going to be an easy solution for that. In other words, there's no way to predict how the positioning is going to be for every person, and this could result in meaningful deviations.
 
Last edited:

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
But if I understand you well, that would result int the same variation in bass response for all headphones ? Sean Olive provided in his presentation the individual measurements for the 20 headphones measured and there were quite a few significant variations in the variation :D.

Headphones like the K550, DT770 pro, etc. showed a significant drop in response below 200-300Hz on the 5128, shaped in a way that's quite similar to seal deterioration, but other closed headphones (Bose, measured passively, for example), didn't exhibit a particularly strong change. Most open headphones such as the HD650, Utopia or HD800 showed a very moderate tilt below 1kHz, but the K701 showed a sharp drop as well.

I'm not certain that a single cause is at play here.

As I'm sure you know, I haven't used either of the rigs (or any rig for that matter, other than my own head and ears sans mic) :). The K550/K553 can be a fairly leaky headphone though, Maya. :) Because it's (apparently) designed for people with larger heads. And has large cups that hang fairly low, which don't seal that well around the bottom or sides on people with smaller heads. And also has rather stiff swivels and hinges that don't naturally adjust to the shape of your head as readily as some other HPs. And has a low clamping force. And because the inserts in the earpads are made of memory foam, that doesn't easily spring back into shape to fill in the gaps between the HPs and your head when you're moving around.

Although I loved my old AKG K553's (which are the updated version of the same design), and used them daily till they died, I had quite a few difficulties with this, even on my rather large-sized noggin. Replacing the memory foam with more springy regular foam inserts definitely helped alot in my case. Adding some extra padding on the headband might also help to improve the position of the earpads on people with smaller heads so they seal somewhat better.

Tyll didn't seem to have much difficulty getting them to seal on his own big head. But he also commented on the seal issues in one of his old reviews, and the difficulties he had getting them to fit and seal properly on his measurement rig...


This is an extremely comfortable headphone, which isolates relatively well, but during measurements I found them finicky about position and clamping force in the extreme. Out of the box, it seems to fit my moderately large head very well, but placing it on my measurement head was another story altogether. People with small heads will need to adjust the headband by pressing it inward and making it curve a bit more to fit tighter and achieve a proper seal. Without doing so will net poor isolation, and a pretty severe change for the worse in sonic character. Once properly adjusted, however, the comfort and sound can be endured for hours with great pleasure.


Some significant changes below 100Hz in the raw frequency response measurements show that these headphones have some difficulty achieving a repeatable seal. I think actual performance is worse than the measurements in this case because I worked fairly hard in achieving a seal each time. The difficulty, I think, is the measurement head is quite a bit smaller than most male heads because it's designed to be an average of both male and female heads, and I think that did come into play with these cans.

Tyll's K550 Graphs: https://www.stereophile.com/images/AKGK550.pdf


I couple photos of my own AKG K553, which show a little better how low the cups can hang on a someone with a somewhat smaller head. These were taken after the headphones died, and the original earpads were in very bad shape, so I just removed them for the photos (which obviously wouldn't help with the sealing issues. :) )

This is with the headband extenders fully retracted...

11282432.jpg


And fully extended...

11282429.jpg
 
Last edited:

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
I have not used the AKG K701. But it has the same size earpads as the K553 (iow, on the larger side). And a different hinge/headband design. I listened to metal571's review though, and he did comment on the foam inserts. And mentioned that they were fairly firm and springy, and did not seem like memory foam, which is probably better from the standpoint of leakage.

metal571 did say that they seemed less bassy than the Q701 though. Which sort of conflicts with Tyll's measurements of the two...


And Tyll's graphs of the K701 also show some variability in the bass, depending on position, like on the K553, which is somewhat interesting...


I believe the K701 (and the Q701), also use angled earpads. So perhaps that contributed to some of the differences in Tyll's bass measurements? And maybe that could also effect how they'd measure on the two rigs, since they would interface a bit differently with the more flat surface on the GRAS than with the more "bumpy" shape of the 5128's mannikin head.
 
Last edited:

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
The other thing that I thought was somewhat interesting in Tyll's review of the Q701 (and K701) was that he said they might need some break in time to sound their best. He specifically mentioned that this might take some of the edginess off of the sound, perhaps implying that maybe that got a bit bassier with wear. And perhaps this could be related to some compression of the earpads.


I tried foam inserts of various thickness with my old AKG K553's. And found that the pad depth seemed to noticeably effect their sound. And the thicker inserts made the headpohones sound a bit brighter and less bassy and warm than the thinner ones. I suspect that this had something to do with either the distance between my ears and the drivers, or possibly the volume of air inside the cups. But not really sure about that. (And I don't have any measurements to back this up.)

 
Last edited:

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
Not unless the majority of people don't like balanced speakers, or they expect a different balance when listening to HPs instead of speakers.
This second possibility is maybe not that farfetched. Didn't Harman determine that was the case? That's why I added the bass preference compensation EQ to the equation, in my proposed method to find a target for the 5128.

Dr. Olive commented not too long ago on this here, sax512. And you can also see my response there as well (along with some examples).


I know this is something that alot of audiophiles probably don't want to hear. But based on looking at the spinorama data of literally dozens of good (and also many not so good) loudspeakers, the bass response on the Harman target seems to track fairly well with the estimated bass levels on neutral loudspeakers that have a built-in sub (or possibly also a separate standalone sub-woofer) which are very well-extended in the sub-bass frequencies. So I guess I don't really see much of a meaningful difference between the two, like some others have described.

One of my hobbies of late has been trying out different loudspeakers' spinorama sound power curves in combination with the 5128 diffuse field compensation curve for use as possible headphone targets. And here's what the estimated sound power responses of three of the best extended neutral loudspeakers look like when combined with the 5128 DF curve...

DUTCH & DUTCH 8C SOUND POWER + 5128 DIFFUSE FIELD

5128PLUSDUTCHDUTCH8C.jpg


INFINITY INTERMEZZO 4.1T SOUND POWER + 5128 DIFFUSE FIELD

5128PLUSINFINITYINTERMEZZO.jpg


KEF REFERENCE 5 SOUND POWER + 5128 DIFFUSE FIELD

5128PLUSKEFREFERENCE5.jpg


This is the average of the three curves above...

5128PLUS3BESTEXTENDED.jpg


And Harman's 2018 over-ear curve for comparison (in blue)...

EX_z4f1U8AA8V6_


The bass levels on the above examples don't look that different to me from the Harman curve.
 
Last edited:

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
924
Likes
1,512
It's pretty typical for different HATS to have variations on this order with a given headphone, particularly those which compress the pinnae somewhat, and thus are particularly sensitive to geometry of the ear both on terms of the air volume presented to the driver, and how the ear tends to fold or bend. In comparisons of a 4128 and KEMAR (both using the same B&K IEC60318-4 ear simulators), I saw comparable or larger variations in that same band on specific headphones.

I've never tried the K550 so can't comment on that, but one thing that appreciate the most with the QC25/35/45 series is the pretty large room inside the cups for the ear lobes.

If I may go back to the question of trying headphones under varying degrees of pad compression in relation to the question of "air volume presented to the driver" (we'll circle back to the GRAS vs. 5128 afterwards), I think that perhaps the QC45 (which in all likelihood is very similar to the QC25/35, the tube linking the front volume to the exterior is in the same place for example, I get the feeling that they recycled most of the design) is an interesting case to study, as :
- the pads are quite thin : there isn't that much compression to be had to start with,
- the pads are designed to "roll" inwards under compression, instead of flat out compressing the foam uniformly, so part of the foam remains uncompressed,
- the room inside means that my ear lobes aren't as much deformed as other HPs of a similar size. Actually they probably aren't.

So while in the case of pad compression it's difficult to separate the influence of various factors from each others (pads side walls deformation, foam compression, ear lobe deformation, change in the volume of air in the front volume), perhaps the QC45 comes the closest of the closed HPs I currently have to reduce the influence of the former factors and express the latter the most.

As already mentioned, even with this sort of moderate compression :
Screenshot 2021-10-02 at 12.24.53.png

I get pretty wild swings in the FR :
Screenshot 2021-10-09 at 09.55.25.png
Same notes and caveats as in this post (with the addition of the H910N and 710BT) : https://www.audiosciencereview.com/...ad-fi-and-sean-olive.27017/page-4#post-927518
Blue traces as they sit on my head, red traces with various degrees of compression / pull.

Perhaps a better way to represent it is to normalise to zero the blue trace and only look at the difference :
Screenshot 2021-10-09 at 10.10.38.png

Another way is to take the above and normalise the differences at, for example, 500Hz, to illustrate how much compression actually changed the overall balance of the headphones :
Screenshot 2021-10-09 at 10.15.55.png
I find the latter graph interesting as it shows that even with much more compression "travel" given its soft and deep pads, and mangling my pinna, the K371 was the least affected of them all (same applies to the 710BT). The SPL was raised fairly uniformly which didn't change that much the "color" of the sound (to be noted is that the blue trace corresponds to me getting a good seal with them, which is not always a given). With a lot less compression travel the ANC headphones were most affected, particularly the QC45 between 1.5 and 4.5kHz (same applies to the H910N).

I no longer have a pair of Bose 700, but I did measure them with various in-ear mics. At the time (early days measuring headphones on-head) I was dubious about the results I was getting with them as while they made sense to a degree in terms of what I was listening, they were so completely off, relatively speaking from other known values such as the HD650, from what the GRAS rigs were obtaining, that I wasn't particularly confident in them.
Low and behold Soundguys measure them on their 5128, and voilà :
Screenshot 2021-10-09 at 10.48.33.png
Mind you the target they used in this graph is absolutely wrong. But using their newer target it gives an indication that it has too much response at around 2kHz or so and 4-5kHz, which is also where I personally tended to want to EQ them down by quite a lot. So at least for me it's a better match (but I wouldn't presume that it is for all).

I find that interesting to compare with the other measurements available of the Bose 700, which show quite wide variation in the ear canal gain range :
- Sean Olive :
Screenshot 2021-09-17 at 08.49.46.png
- Oratory's preliminary PDF :
Screenshot 2021-10-09 at 10.52.08.png
- Oratory's current PDF :
Screenshot 2021-10-09 at 10.52.14.png
- Crinacle : https://crinacle.com/graphs/headphones/bose-noise-cancelling-headphones-700/
- And of course ASR : https://www.audiosciencereview.com/...c700-review-noise-cancelling-headphone.24667/

I suspect that the 700 shares a fairly similar behaviour as the QC45 under compression (same goes for the Sony H910N, cf. previous post, and I suspect most of the high-performance ANC models, with the Airpods Max being perhaps less affected).

I've also noticed that both the QC45 and H910N have a "zero" point where the compression doesn't change the SPL at all, even passively. Below that point they behave under compression in a fairly moderate, smooth way, above that point they swing quite wildly up to 5kHz or so. Coincidentally or not on the QC45 that "zero" point is between 1 and 2kHz, only a little bit above where the ANC feedback stops operating.

If we put aside the questions of pad side walls deformation, foam compression, ear lobe deformation (we shouldn't but let's accept that fiction for a moment), and only focus on volume of air presented to the driver, what mechanism would be at play to explain this behaviour, particularly for ANC headphones (even passively, cf. graphs for the QC45 and H910N) ?

Oratory mentioned somewhere (I believe on discord) that ANC headphones have to be designed a specific way in terms of source acoustic impedance to perform well. Is their more sensitive behaviour in the 1-5kHz range, even measured passively, a compromise derived from their specific needs for ANC ? Could even a moderate variation in the volume of air in the front volume realistically result in such wide variations ?

Could this potentially mean that small discrepancies between test rigs and the average human, or a singular individual, that didn't matter quite as much before to still provide a good representation of what the actual experience would be, may now start to matter with such headphones ?
 
Last edited:
D

Deleted member 16543

Guest
Are you suggesting that there are issues with B&K's underpinning research? It seems to agree well with a randomly grabbed paper from an unrelated source, and, of course, underpins the present ITU-T standards for standard human ear Z - I'd expect that the working group at the ITU would have raised an issue if there were an issue...

I think his position is more along the lines of 'theory is all nice and dandy, but I can't practically use the rig without a target curve, so it's useless to me'.

Being a trained engineer I can't possibly not appreciate his point of view. If we were to wait for perfection, we'd still be fighting over which ficus leaf shape best covers our groin. But..

You can't possibly become an engineer without having been exposed to physics, and I personally believe that you can't possibly become the best engineer you can be without loving physics. And physics says what it says about geometric boundaries, for the solution of the wave equation. And that gives the 5128 the edge. Incontrovertibly.

I'll dig through past communications and try to find the study I was referring to about the flat wave front behavior engaging point inside the canal as function of frequency.
I vaguely remember a pic that shows turbulence way down the canal, at frequencies that most people still can hear.
It's possible It wasn't you I was talking to about that, though.

As for the study from the mid 90's, that's exactly the one I was referring to for proof that directivity matters down to about 3 kHz at least, when comparing mic capsule at the CLOSED canal entrance vs eardrum position.
That doesn't mean it differs by the same amount when comparing the waves at the OPEN canal entrance vs the eardrum, which is what we need to ascertain the superiority of the B&K vs. GRAS rig, but it does mean that there is directivity information at least up to that point, and placing a capsule at the canal entrance distorts it.
And if that were the case only up to that point, with the canal representing a fixed added impedance load to the eardrum, for all frequencies, why the impossibility to transpose the GRAS target curve to the B&K rig? If canal shape didn't matter, why couldn't we find a simple, univocally determined EQ to go from one to the other and vice versa?
Also, more philosophically, speaking, why would nature bother to create a canal in the first place? I can't think of one instance where something in nature or biology isn't optimized to the max, so there has to be a function for the canal, otherwise, through evolution, we would not have it anymore by now. But I'm no biologist and I digress..

I know you ascribed the deviations in that study to positioning errors, but the person at DIYaudio who took some of these measurements again with his own rig (Baffless was his handle) paying a lot of attention to minimizing that risk (unavoidably without being able to do it 100%, of course) found similar results.

In my opinion, the question 'how far can we deviate from the shape of an anthropomorphic canal to get results that are still valuable, at least up to a certain margin/frequency?', considering the new available choices in rigs, needs to be replaced with the question 'now that we don't have to worry about that, how do we move forward?'.
And the answer to that is by providing a target curve for the more realistic binaural mics that are already in the market and will eventually come out in the future.
It's hard to imagine, for example, that KEMAR will be that much far behind with an anatomically accurate ear option as one of their products.
I think the head start will go to whoever will provide a completely anthropomorphic dummy head with its specific target curve.
 
Last edited by a moderator:

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
I've never tried the K550 so can't comment on that, but one thing that appreciate the most with the QC25/35/45 series is the pretty large room inside the cups for the ear lobes.

If I may go back to the question of trying headphones under varying degrees of pad compression in relation to the question of "air volume presented to the driver" (we'll circle back to the GRAS vs. 5128 afterwards), I think that perhaps the QC45 (which in all likelihood is very similar to the QC25/35, the tube linking the front volume to the exterior is in the same place for example, I get the feeling that they recycled most of the design) is an interesting case to study, as :
- the pads are quite thin : there isn't that much compression to be had to start with,
- the pads are designed to "roll" inwards under compression, instead of flat out compressing the foam uniformly, so part of the foam remains uncompressed,
- the room inside means that my ear lobes aren't as much deformed as other HPs of a similar size. Actually they probably aren't.

So while in the case of pad compression it's difficult to separate the influence of various factors from each others (pads side walls deformation, foam compression, ear lobe deformation, change in the volume of air in the front volume), perhaps the QC45 comes the closest of the closed HPs I currently have to reduce the influence of the former factors and express the latter the most.

As already mentioned, even with this sort of moderate compression :
View attachment 158087

I get pretty wild swings in the FR :
View attachment 158089
Same notes and caveats as in this post (with the addition of the H910N and 710BT) : https://www.audiosciencereview.com/...ad-fi-and-sean-olive.27017/page-4#post-927518
Blue traces as they sit on my head, red traces with various degrees of compression / pull.

Perhaps a better way to represent it is to normalise to zero the blue trace and only look at the difference :
View attachment 158090

Another way is to take the above and normalise the differences at, for example, 500Hz, to illustrate how much compression actually changed the overall balance of the headphones :
View attachment 158092
I find the latter graph interesting as it shows that even with much more compression "travel" given its soft and deep pads, and mangling my pinna, the K371 was the least affected of them all (same applies to the 710BT). The SPL was raised fairly uniformly which didn't change that much the "color" of the sound (to be noted is that the blue trace corresponds to me getting a good seal with them, which is not always a given). With a lot less compression travel the ANC headphones were most affected, particularly the QC45 between 1.5 and 4.5kHz (same applies to the H910N).

I no longer have a pair of Bose 700, but I did measure them with various in-ear mics. At the time (early days measuring headphones on-head) I was dubious about the results I was getting with them as while they made sense to a degree in terms of what I was listening, they were so completely off, relatively speaking from other known values such as the HD650, from what the GRAS rigs were obtaining, that I wasn't particularly confident in them.
Low and behold Soundguys measure them on their 5128, and voilà :
View attachment 158093
Mind you the target they used in this graph is absolutely wrong. But using their newer target it gives an indication that it has too much response at around 2kHz or so and 4-5kHz, which is also where I personally tended to want to EQ them down by quite a lot. So at least for me it's a better match (but I wouldn't presume that it is for all).

I find that interesting to compare with the other measurements available of the Bose 700, which show quite wide variation in the ear canal gain range :
- Sean Olive :
View attachment 158094
- Oratory's preliminary PDF :
View attachment 158095
- Oratory's current PDF :
View attachment 158096
- Crinacle : https://crinacle.com/graphs/headphones/bose-noise-cancelling-headphones-700/
- And of course ASR : https://www.audiosciencereview.com/...c700-review-noise-cancelling-headphone.24667/

I suspect that the 700 shares a fairly similar behaviour as the QC45 under compression (same goes for the Sony H910N, cf. previous post, and I suspect most of the high-performance ANC models, with the Airpods Max being perhaps less affected).

I've also noticed that both the QC45 and H910N have a "zero" point where the compression doesn't change the SPL at all, even passively. Below that point they behave under compression in a fairly moderate, smooth way, above that point they swing quite wildly up to 5kHz or so. Coincidentally or not on the QC45 that "zero" point is between 1 and 2kHz, only a little bit above where the ANC feedback stops operating.

If we put aside the questions of pad side walls deformation, foam compression, ear lobe deformation (we shouldn't but let's accept that fiction for a moment), and only focus on volume of air presented to the driver, what mechanism would be at play to explain this behaviour, particularly for ANC headphones (even passively, cf. graphs for the QC45 and H910N) ?

Oratory mentioned somewhere (I believe on discord) that ANC headphones have to be designed a specific way in terms of source acoustic impedance to perform well. Is their more sensitive behaviour in the 1-5kHz range, even measured passively, a compromise derived from their specific needs for ANC ? Could even a moderate variation in the volume of air in the front volume realistically result in such wide variations ?

Could this potentially mean that small discrepancies between test rigs and the average human, or a singular individual, that didn't matter quite as much before to still provide a good representation of what the actual experience would be, may now start to matter with such headphones ?

I don't know if this would explain the differences in the measurements between the GRAS and 5128. But I think that both pad depth and compression could potentially effect a headphone's sound. And they might effect its sound differently when measured on a fixture with a simulated ear attached to a more or less flat surface, like on the GRAS 45C...

index.php


...versus a more bumpy and rounded surface, like on the 5128 mannikin heads.

11220116.png


Assuming that you are using roughly the same type and shape of artificial ears on both setups, you would think that the drivers in a larger circumaural/over-ear headphone would probably be seated a little more closely to the opening of the ear canal on the more curved surface of the mannikin, than on a flat plate (because the area around the ear on the mannikin head is a bit more recessed). Which would seem to favor a more bassy response on the mannikin, in my experience. Rather than the other way around.

The more curved/bumpy surface of the mannikin head probably also has more opportunities for the headphone to leak though. And if the pads on the headphone are more compressed, they may not fill in the gaps between the simulated head and headphones quite as well. (I think Dr. Olive mentioned the cheek as one possible area of leakage on the 5128 mannikin head. And the fixture does appear to be curved/sloped fairly forward in that area in the above picture. Though it's a bit hard to tell from this particular POV. How realistic that may be though for a human head, I couldn't really say.)

IAC, a circumaural headphone is almost certainly going to interface/couple a bit differently with these two types of fixtures. Which will probably contribute to some differences in the measurements between the two.
 
Last edited:

DualTriode

Addicted to Fun and Learning
Joined
Oct 24, 2019
Messages
893
Likes
593
Hello All,

A 5128 may model a theoretical ear better than an ear trumpet.

So What!

If you are building a gold standard measurement tool from scratch, leave the artificial couplers and pinna out of it. They are not required. You have your own ears.

Build a test “head” flat on both sides where the ears would be and drill holes to accept GRAS lab quality 46AO 1/2'' CCP Pressure Standard Microphones.

Now place your favorite headphones on the flat world test fixture. Run your software test sweep a few times and plot the curve. This new curve is your new “standard target curve”.

Place another headphone on your new “flat world” test fixture. Run a test sweep with your software. Construct a transfer function in your software to match this new Headphone Frequency Response to the “new standard target curve”. Now this second set of headphones will sound much like your original favorite set of headphones.

No perfect set of artificial couplers or pinna need apply or are required, leave them out of the process.

To personalize your headphone listening you may want to add your own personal correction curve to the new “standard target curve”.

Thanks DT

Edit:
@solderdude may have the correct idea.
 
Last edited:

Mad_Economist

Addicted to Fun and Learning
Audio Company
Joined
Nov 29, 2017
Messages
543
Likes
1,618
Could this potentially mean that small discrepancies between test rigs and the average human, or a singular individual, that didn't matter quite as much before to still provide a good representation of what the actual experience would be, may now start to matter with such headphones ?
Great post overall, I was a little overwhelmed when I got the notification, but the data is interesting.

To the question, it seems conceivable to me...but bear in mind that there isn't a difference in ear Z between the 5128 and 45CA - or any IEC60318-7 compliant HATS - in the band you're talking about. The ITU-T P57 4.3 ear's Z differs at low and high frequency extremes, but generally matches the older 3.3 standard in the midrange and low to mid treble.

I think his position is more along the lines of 'theory is all nice and dandy, but I can't practically use the rig without a target curve, so it's useless to me'.
I will point out, it isn't that hard to make a target for the 5128, but yes, if you're looking for something in the vein of the Harman work, there is no better alternative to the 45CA and its GRAS family for the moment (which is why, memory serving, I recommended it over the 5128 some time ago).

Also, more philosophically, speaking, why would nature bother to create a canal in the first place? I can't think of one instance where something in nature or biology isn't optimized to the max, so there has to be a function for the canal, otherwise, through evolution, we would not have it anymore by now. But I'm no biologist and I digress..
If I may point you to the middle ear, a frankly rather awful bandpass required simply because our cochlea does not function without a water-like fluid filling but our ears are coupled to air, as an example of a sub-optimal element of ear design...

More broadly, the canal has a number of obvious functions: it protects the tympanum and is part of what provides the boost in our most sensitive audible frequencies. This doesn't mean it plays a role in directional audio, which is a point of view which, as far as I can tell, is entirely unique to you, specifically.

I know you ascribed the deviations in that study to positioning errors, but the person at DIYaudio who took some of these measurements again with his own rig (Baffless was his handle) paying a lot of attention to minimizing that risk (unavoidably without being able to do it 100%, of course) found similar results.
If you believe that there's a consistently demonstrable directional effect of the canal that low in frequency (3khz), and that Baffless' methodology will demonstrate it, I strongly suggest you publish on it, because it would be novel science as far as I can tell. I've read Baffless' posts and I don't find them to compellingly demonstrate that he's controlled for by far the most obvious source of variation here, and the fact that the simultaneous in situ measurements in Middlebrooks et al doesn't show anything there further affirms that position.

Place another headphone on your new “flat world” test fixture. Run a test sweep with your software. Construct a transfer function in your software to match this new Headphone Frequency Response to the “new standard target curve”. Now this second set of headphones will sound much like your original favorite set of headphones.
I've posted about this before, but you cannot extrapolate from an earless system to an eardrum response with strong confidence.
 

metal571

Member
Forum Donor
Joined
Dec 1, 2017
Messages
11
Likes
52
I have not used the AKG K701. But it has the same size earpads as the K553 (iow, on the larger side). And a different hinge/headband design. I listened to metal571's review though, and he did comment on the foam inserts. And mentioned that they were fairly firm and springy, and did not seem like memory foam, which is probably better from the standpoint of leakage.

metal571 did say that they seemed less bassy than the Q701 though. Which sort of conflicts with Tyll's measurements of the two...


And Tyll's graphs of the K701 also show some variability in the bass, depending on position, like on the K553, which is somewhat interesting...


I believe the K701 (and the Q701), also use angled earpads. So perhaps that contributed to some of the differences in Tyll's bass measurements? And maybe that could also effect how they'd measure on the two rigs, since they would interface a bit differently with the more flat surface on the GRAS than with the more "bumpy" shape of the 5128's mannikin head.
Depends on which K701 it is. I specifically sought an early Austrian made K701. Changes to production over the years and use of different foam under the pads are highly likely. I know for absolutely certain that what my ears heard between my Chinese made Q701 and my Austrian K701 was a noticeable difference in bass quantity. Graphs are graphs of a *single* unit. They cannot represent every revision made to a model that has been manufactured for over a decade now.
 
D

Deleted member 16543

Guest
Great post overall, I was a little overwhelmed when I got the notification, but the data is interesting.

To the question, it seems conceivable to me...but bear in mind that there isn't a difference in ear Z between the 5128 and 45CA - or any IEC60318-7 compliant HATS - in the band you're talking about. The ITU-T P57 4.3 ear's Z differs at low and high frequency extremes, but generally matches the older 3.3 standard in the midrange and low to mid treble.


I will point out, it isn't that hard to make a target for the 5128, but yes, if you're looking for something in the vein of the Harman work, there is no better alternative to the 45CA and its GRAS family for the moment (which is why, memory serving, I recommended it over the 5128 some time ago).


If I may point you to the middle ear, a frankly rather awful bandpass required simply because our cochlea does not function without a water-like fluid filling but our ears are coupled to air, as an example of a sub-optimal element of ear design...

More broadly, the canal has a number of obvious functions: it protects the tympanum and is part of what provides the boost in our most sensitive audible frequencies. This doesn't mean it plays a role in directional audio, which is a point of view which, as far as I can tell, is entirely unique to you, specifically.


If you believe that there's a consistently demonstrable directional effect of the canal that low in frequency (3khz), and that Baffless' methodology will demonstrate it, I strongly suggest you publish on it, because it would be novel science as far as I can tell. I've read Baffless' posts and I don't find them to compellingly demonstrate that he's controlled for by far the most obvious source of variation here, and the fact that the simultaneous in situ measurements in Middlebrooks et al doesn't show anything there further affirms that position.


I've posted about this before, but you cannot extrapolate from an earless system to an eardrum response with strong confidence.

I don't have to demonstrate it. It has been measured! And also modeled (I know.. I have to find that pic I'm talking about).
And speaking about the fluid in the middle ear, that also serves the purpose of maintaining your balance. Pretty optimized, if you ask me. But obviously this is no substitute for a mathematical demonstration.

Still, my question stands: why not a univocal transfer function between the two rigs? At least for iems, since we seem to at least agree on the fact that the pinna imprints directional information to the sound waves hitting it.
 

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
6,878
Likes
6,674
Location
UK
Depends on which K701 it is. I specifically sought an early Austrian made K701. Changes to production over the years and use of different foam under the pads are highly likely. I know for absolutely certain that what my ears heard between my Chinese made Q701 and my Austrian K701 was a noticeable difference in bass quantity. Graphs are graphs of a *single* unit. They cannot represent every revision made to a model that has been manufactured for over a decade now.
Pad wear can make a big difference on headphones, for instance this is what Oratory measured on one of my K702's with old vs new pads on the same headphone, he said to imagine the red line measurement was inaccurate below around 60Hz and it instead would track the same difference to the green line without crossing it (a measurement error below 60Hz for the red line), so it's basically a 2dB Low Shelf from 1000Hz, so the older pads make it sound warmer (more bass) which would be similar for your AKG headphones you're talking about, so you'd have to take pad wear into consideration between the two headphones (and anyway Q701 and K701 are not the same headphone so not sure how you can compare bass levels in terms of concluding Chinese vs Austrian differences):
K702 old pads vs new pads.png

There also can be quite a lot of unit to unit variance between at least the K702 headphones, so that's another variable that could affect your Chinese vs Austrian AKG comparisons - ie make it less valid potentially, following is the frequency response of my two Chinese K702's with new pads:
#2 K702 channel matching.png

So I think it's not easy for you to drill down to say it's a Chinese vs Austrian thing or even how the headphone specifications may or may not have changed through the years - there's too much variation due to pad wear & unit to unit variation to draw such fine conclusions in my view.

But some of what I'm saying here might be relevant for what @ADU is observing in the discrepancies of measurement he's pointed out.
 
Last edited:

DualTriode

Addicted to Fun and Learning
Joined
Oct 24, 2019
Messages
893
Likes
593
Hello @Mad_Economist ,

We are not speaking of an earless system. We have our own ears.

I read your previous posts and the linked AES Floyd E, Toole paper, “THE ACOUSTIC AND PSYCHOACOUSTICS OF HEADPHONES”. My impression from the Floyd E, Toole paper that the artificial ear becomes more important with the addition of variables beyond Frequency Response alone.

Let me point out that when we are talking about target curves we are speaking of Frequency Response and transfer functions not spatialization or other variables. Harman research tells us that the most important variable is Frequency Response.

I do not believe that the Harman Target Curve addresses anything other than Frequency Response. An artificial ear need not apply. We do not need a HATS, we are not creating 3D artificial space.

Thanks DT

If this is not fun I will stop.
 
Top Bottom