[ADUHF]

Someone in one of the livestreams was asking if Stereophile would repost Tyll's old articles and reviews from Inner Fidelity btw. Many of his articles seem to have gotten lost during the dissolution of Inner Fidelity. Including, regrettably, his articles on the Harman target. I believe all of his headphone-related reviews and measurements can still be found here though, in case anyone is interested...

https://www.stereophile.com/writer/15081

 

[MayaTlab]

I apologize if this is something you already mentioned elsewhere, because I'm not immediately seeing it above. But when you do your own measurements, do you take them with the headphones at several different positions on your head, like Tyll used to do, and then compute an average of those? Or do you just try to get the seal and and fit as correct as possible based on your own listening, and then do just a single measurement from that position?

Depends on the intended application. Single measurements may occasionally provide a sharper illustration of high-Q problems than averages. If I want to test EQ vs no EQ or different EQ profiles vs each others for the same headphones it's logical to not move the headphones between the measurements. If I want to look at the seatings to seatings spread I'll simply display the bunch of individual measurements taken during the same session.
But since I'm egoistically doing these measurements to improve my listening experience with EQ the thing I most often look at is averages of between 5 - 10 realistic seatings on my head (ie taking the headphones off and putting them back on as I would in a normal situation). On some HPs it gets tedious when after having done your 10 measurements you end up with lines nearly perfectly landing on each others, as in a "why bother other than in the name of rigour" way (my HD650 with dekoni pads - stock pads are slightly less consistent). On others you quickly realise how extremely important averaging is to even start to get an idea of the FR curve you can use to base an EQ profile on. More importantly I rinse and repeat the process several times over days, weeks, now months...
I'll occasionally plot averages of averages to make some trends appear in a sharper way, and I'll frequently plot measurement sessions or different mics against each others to assess the repeatability of the absolute or relative values and the existence or not of a constant transfer function between them or not.
I've done a few tests for unrealistic seatings (too far backwards, upwards, etc.), but that's more for curiosity's sake than anything else.
I also occasionally measure headphones for consistency when tilting my head forwards / backwards / left / right.
Basically it really depends on what I want to do with the measurements and then I work my way backwards.

Measuring headphones on one's head is different in that regard from measuring headphones on a HATS I think. On your head there's just only so much positions you can try before it starts to intuitively feel wrong.
Besides, when you measure on a HATS, particularly for EQ purposes, as Oratory does, you do so to share your results with others. I think that spatial averaging makes sense to remove / soften the peaks / dips that are a product of a specific location and only keep the ones that are constant. You may also want to present various case scenarios for the FR at lower frequencies you get depending on seal quality, something which on your own head is a by-product of where HPs will naturally sit that you may be forced to accept (you may not wear a pair of headphones in a location that ensures a better seal if they're uncomfortable to wear this way, this sort of "take it or leave it" thing). Basically, HATS measurements and individual in-ear measurements necessarily serve different purposes and may require different methods to produce results relevant to the intended application.

You should try your hand at in-ear measurements, at least between 50-800Hz for relative comparisons between headphones it's fairly straightforward and not that expensive.

 

[ADUHF]

Thank you for clarifying the above, Maya. One other question, or comment...

You mentioned "nulls" at certain frequencies on a couple of the headphones. I assume you're not referring to the normal resonant peaks and dips that show up fairly consistently in the same places in the treble measurements of all three headphones for your ears.

You should try your hand at in-ear measurements, at least between 50-800Hz for relative comparisons between headphones it's fairly straightforward and not that expensive.

It's an interesting idea.

If I did something like this, then I'd also want to do in-ear measurements of loudspeakers for comparison with the headphones. Resolve suggested he might try something like this. And this is something I have also advocated, for quite a long time. And continue to suggest to the other measurement and headphone review sites, and the manufacturers of their measurement rigs, to help users of their measurements to get a better idea how the headphone measurements on their rigs compare to loudspeakers.

It would obviously take some work and experimentation to do well. But if you're a review site (or HATS mfr.), and not interested in doing that kind of work to better aid and inform your readers/users, and to validate some of your conclusions, then maybe doing headphone reviews and transducer measurements isn't really your thing? Just my 2c fwiw. That does not mean that I am not grateful for the info that some of these sites provide though. Because I do find some of it to be pretty useful in my buying and EQ-ing decisions. I just wish it could be even more useful!

I'd be curious to see some of your in-ear speaker measurements as well, for comparison with your headphone measurements, to see how close or far away they are.

I also noticed in one of your graphs that you did some measurements of the Sennheiser HD560S. Which is also one of the open-back headphones I've been considering. There are two things that concern me though about its frequency response. The first and most important would be the possible imbalance in the upper mids that appears on this Rtings FR plot at around 2k, where the right channel appears to dip noticeably lower than the left channel...

https://www.rtings.com/headphones/1-5/graph#18492/7903

Most of the Senn 5 and 6 series graphs I've looked at have had some fairly noticeable imbalances in the treble region. So I assume this is not just a measurement blip on the 560S. I listened to reviews of the 560S by several prominent YT reviewers though. Some of which were more complimentary than others. Only one seemed to be noticeably bothered though by this particular imbalance...



The other thing that concerns me a bit less is the 560S's potential brightness in some spots in the upper mids or low treble. Because my other headphones, the DT770 and M50x, already have similar issues. If the HD560S is noticeably on the brighter side, then that is probably something which could be easily corrected with an EQ. The possible imbalance in the upper mids could potentially be more difficult to correct though. (If it is indeed there and noticeable.)

I'm also somewhat hesitant to invest in another dynamic headphone before trying a planar magnetic, since I already have two other pretty good dynamic HPs. It sounds like the newer polymer drivers and open design of the HD560S might have a few advantages though versus the closed dynamic headphones I've used up to now.

Imo, the AKG K553, 250-ohm Beyer DT-770, and AudioTechnicas M50x can all produce some fairly decent results with a bit of EQ, and L/R balance adjustment. And with these sort of adjustments, I'd put them pretty close to the top of the heap in their price range.

Prices have been steadily dropping on open-back and planar magnetic headphones though in the last several years. And there are now at least a couple of halfway decent models which are in the same price ranges as the closed headphones above. So it is not as easy to simply write them off anymore, because they're either too expensive, or not well-extended in the bass. If you're looking in the $150-$250 price range, I think you have to at least be open to the idea of something different than some of the older and (previously) more affordable closed-backs designs. Some of which have been around for literally decades.

 

[MayaTlab]

If I did something like this, then I'd also want to do in-ear measurements of loudspeakers for comparison with the headphones. Resolve suggested he might try something like this. And this is something I have also advocated, for quite a long time. And continue to suggest to the other measurement and headphone review sites, and the manufacturers of their measurement rigs, to help users of their measurements to get a better idea how the headphone measurements on their rigs compare to loudspeakers.

It would obviously take alot of work and some experimentation to do well. But if you're a review site, and not interested in doing that kind of work to better aid and inform your readers/users, and to validate some of your conclusions, then perhaps doing headphone reviews and transducer measurements isn't really your bag. And you should be thinking about looking for work in some other type of occupation... Just my 2c fwiw.

I'd be curious to see some of of your in-ear speaker measurements though, for comparison with your headphone measurements to see how close, or far away they are.

This is a lot more difficult to do and interpret well than what it seems at first.

The first problem is to know what you want to do with these measurements. Are you trying to mimic your own speakers in your own room (Smyth Realizer, Impulcifier) ? Are you trying to measure your own HRTF ? Are you trying to "do a Harman" ? Etc. Some of these things may be practically possible, some (most) won't.

The second problem is to make sure that the microphone system you'll be using is fit for the purpose. Even with the "simplest" thing you can do (Smyth Realiser, Impulcifier) at very least you'll need ear canal entrance measurements fit for relative measurements up to 10kHz... and I'm already not satisfied by the fact that I'm not finding, with headphones, a superbly constant transfer function between my blocked ear canal measurements (inserted as deeply as I could - even a slight protrusion from the entrance ruins the transfer function, at least for me) and my probe mic past 7kHz or so. It's not that bad, but not that satisfyingly matching either :

Difference between the measurements from my pair of HD650 with Dekoni Elite Velour pads and three other headphones (HD560S, H400SE, Hi-X65), normalised at 500Hz, right ear, for one session each for the probe mic (1 in the image above, solid traces), blocked canal measurements, inserted as deep as I could with the apparatus I used (2, dotted traces), and the same but protruding slightly from the canal entrance (3, small dots). The HD650, HD560S and H400SE have fairly low seatings variation, the Hi-X65 less so (so don't nitpick as much about their results).
In general, taking into consideration the Hi-X65 slightly higher seatings to seatings variation in the range of interest, I tend to find a pretty constant transfer function between 1) and 2) up to around 1600Hz and between around 4-7kHz.
The differences in the ear canal gain region can be explained by comparing 1) and 2) with in concha measurements with the ear canal open (4) and blocked (5). All blocked canal measurements (2, 5) agree very well with each others in the ear canal gain region. The open canal measurements (1, 4) don't, BUT, with great constance, the in-concha measurements with open canal are always the closest trace to the probe mic and both are always veering in the same direction relative to the blocked canal measurements, so I'm tempted to take a leap of faith and presume that the "truth" lays somewhere in the direction both indicate. It's interesting to note that open canal measurements don't diverge from blocked canal measurements in a constant fashion : the H400SE and HD560S traces go in opposite directions for example.
Above 7kHz I'm not getting a superb match between 1) and 2). It's even worse if the blocked ear canal measurement apparatus protrudes even slightly from the canal entrance, the lack of constant transfer function starts to appear at lower frequencies (3, small dotted traces). I'll try to re-design my blocked ear canal measurement apparatus to insert it even deeper. In general the blocked ear canal measurements with "deep" insertion (2) I've done so far tend to have better repeatability than the probe mics for absolute values, but less so for relative comparisons between headphones, and the blocked ear canal measurements relative results seem highly susceptible to their exact position relative to the ear canal entrance.
Now, I have a pretty good idea which one is more representative of what I'm hearing in the 7-10kHz range, but the reason why is less satisfying and more subjective.
TLDR : as soon as your in-ear measurements ambitions go beyond measuring headphones against each others below 800Hz, things get suddenly a lot more complicated.
I haven't compared these measurement techniques with speakers. Will I get similar tendencies or not ? No idea.

Personally I'm not comfortable with my own in-ear speakers measurements beyond just using them as rough indicators to increase or decrease my confidence in the way I nudged my EQ profiles one way or another, a posteriori. I mean, I did indeed measure my near-fields in very near field conditions with my probe mics at various angles relative to my head, but what do the results I'm getting tell me ? I'm not actually certain of that.

More useful IMO is to make sure that your speaker system is set up as well as you can, and just listen to music with it vs. headphones . A good pair of near-fields decently set up remains to me the "reference". For me that was enough to tell me a while ago that, at least as far as I'm concerned, unEQed headphones suck .

 

[castleofargh]

I think Resolve also referenced some recent remarks by Steve Guttenberg in one of his recent live streams as well. And that may have been how the subject of measurements, and whether they actually matter came up. So here is also Mr. Guttenberg, with a somewhat different perspective than the above.

Guttenberg is the kind of person I would love to have as IRL friend(that's just how much I respect and like the guy in general). I also found myself sharing his subjective taste for gears often enough over the years. But I wouldn't put him in a panel to discuss measurements or their interpretation.

Tool on the other hand, IDK if he's fun at parties, but he's been as solid as it gets on speaker research. So anytime their views diverge about measurements, you know where my vote goes. On more subjective stuff, it's another story.

You should try your hand at in-ear measurements, at least for between 50-800Hz for relative comparisons between headphones it's fairly straightforward and not that expensive.

I told him already. It's written all over his posts that he's made for that stuff. And he often enough gets confused by a lack of personal experience with measurements. ”This is the way”.

 

[MayaTlab]

I told him already. It's written all over his posts that he's made for that stuff.

Haha let's make it happen then .
@ADUHF feel free to ask how to get started if I can help.

 

[ADUHF]

This is a lot more difficult to do and interpret well than what it seems at first.

The first problem is to know what you want to do with these measurements. Are you trying to mimic your own speakers in your own room (Smyth Realizer, Impulcifier) ? Are you trying to measure your own HRTF ? Are you trying to "do a Harman" ? Etc. Some of these things may be practically possible, some (most) won't.

Thank you for your detailed reply on this Maya.

I'm afraid that most of the things I'd want to do with measurements would not be practical on my budget. Because I don't have any personal fortunes lying around anywhere at the moment. When it comes to the tonal balance and frequency response of headphones, there is really only one thing that I want to know at the moment. And that is-- how close is it to the sound of anechoically flat loudspeakers in a typical domestic listening space?... That's really about it.

There are others things it could potentially be useful for, like equalization, checking driver symmetry, looking at the effects of physical modifications, and so forth. But comparing the sound to a neutral reference (which equals anechoically flat loudspeakers in a room imho), is really the thing that I'm most interested in.

There probably isn't just one solution to this either. Because different speakers have different dispersion characteristics, bass extension, and so forth. And those variations would needed to be accounted for in some way, probably by averaging the responses of several different loudspeakers. And there are also other factors that should probably be taken into account, such as the reflectivity, shape and contents of the room. The position and biometrics of the listener, and so forth. For someone who has the time, the money and the desire though. That is all very doable. And it doesn't require any subjective tests.

There is alot that I can do though along these lines with just the sound power measurements of loudspeakers. And diffuse field measurements of headphones... thanks to graphers like Oratory1990, Amir and Pierre Aubert at ASR, and other folks like that. I've been wanting do some comparisons using Jude's HBK 5128 measurements as well. And it looks like that may also be possible using the diffuse field compensation curve for the 5128 rig on page 2 of this brochure...

https://www.bksv.com/-/media/literature/Product-Data/bp2573.ashx

It looks like the DF and Free-field curves could be tailored specifically for each individual unit of the 5128 that they sell. And the curve shown in the above PDF is just an average. Given how carefully HBK manufactures these things though, an average DF curve may be more than good enough. (And certainly better than nothing at all.)

The DF plot in the above PDF is for the 5128 system with the torso btw. Which is probably preferable to a DF curve for just the head, since it incorporates some of the body geometry in the DF measurements as well.

The second problem is to make sure that the microphone system you'll be using is fit for the purpose. Even with the "simplest" thing you can do (Smyth Realiser, Impulcifier) at very least you'll need ear canal entrance measurements fit for relative measurements up to 10kHz... and I'm already not satisfied by the fact that I'm not finding, with headphones, a superbly constant transfer function between my blocked ear canal measurements (inserted as deeply as I could - even a slight protrusion from the entrance ruins the transfer function, at least for me) and my probe mic past 7kHz or so. It's not that bad, but not that satisfyingly matching either :


Difference between the measurements from my pair of HD650 with Dekoni Elite Velour pads and three other headphones (HD560S, H400SE, Hi-X65), normalised at 500Hz, right ear, for one session each for the probe mic (1 in the image above, solid traces), blocked canal measurements, inserted as deep as I could with the apparatus I used (2, dotted traces), and the same but protruding slightly from the canal entrance (3, small dots). The HD650, HD560S and H400SE have fairly low seatings variation, the Hi-X65 less so (so don't nitpick as much about their results).
In general, taking into consideration the Hi-X65 slightly higher seatings to seatings variation in the range of interest, I tend to find a pretty constant transfer function between 1) and 2) up to around 1600Hz and between around 4-7kHz.
The differences in the ear canal gain region can be explained by comparing 1) and 2) with in concha measurements with the ear canal open (4) and blocked (5). All blocked canal measurements (2, 5) agree very well with each others in the ear canal gain region. The open canal measurements (1, 4) don't, BUT, with great constance, the in-concha measurements with open canal are always the closest trace to the probe mic and both are always veering in the same direction relative to the blocked canal measurements, so I'm tempted to take a leap of faith and presume that the "truth" lays somewhere in the direction both indicate. It's interesting to note that open canal measurements don't diverge from blocked canal measurements in a constant fashion : the H400SE and HD560S traces go in opposite directions for example.
Above 7kHz I'm not getting a superb match between 1) and 2). It's even worse if the blocked ear canal measurement apparatus protrudes even slightly from the canal entrance, the lack of constant transfer function starts to appear at lower frequencies (3, small dotted traces). I'll try to re-design my blocked ear canal measurement apparatus to insert it even deeper. In general the blocked ear canal measurements with "deep" insertion (2) I've done so far tend to have better repeatability than the probe mics for absolute values, but less so for relative comparisons between headphones, and the blocked ear canal measurements relative results seem highly susceptible to their exact position relative to the ear canal entrance.
Now, I have a pretty good idea which one is more representative of what I'm hearing in the 7-10kHz range, but the reason why is less satisfying and more subjective.
TLDR : as soon as your in-ear measurements ambitions go beyond measuring headphones against each others below 800Hz, things get suddenly a lot more complicated.
I haven't compared these measurement techniques with speakers. Will I get similar tendencies or not ? No idea.

I'm not sure I'm following all of the above. The plots of the 3 headphones versus the HD650 are interesting though.

Personally I'm not comfortable with my own in-ear speakers measurements beyond just using them as rough indicators to increase or decrease my confidence in the way I nudged my EQ profiles one way or another, a posteriori. I mean, I did indeed measure my near-fields in very near field conditions with my probe mics at various angles relative to my head, but what do the results I'm getting tell me ? I'm not actually certain of that.

More useful IMO is to make sure that your speaker system is set up as well as you can, and just listen to music with it vs. headphones . A good pair of near-fields decently set up remains to me the "reference". For me that was enough to tell me a while ago that, at least as far as I'm concerned, unEQed headphones suck .

This is pretty much the conclusion I reached as well. At least for most of the headphones that are currently within my grasp and price range.

I'm still sort of curious to see your in-ear measurements from the speakers though. (Even if they aren't particularly good.) And to compare them to the raw headphone plots on the same graph, if that's feasible. Because I think that would be very interesting and educational.

If more users of this kind of measurement gear did this, then maybe some of the bigger headphone review sites (like Rtings) would also want to do it. And provide this sort of info to their users.

 

[ADUHF]

Guttenberg is the kind of person I would love to have as IRL friend(that's just how much I respect and like the guy in general). I also found myself sharing his subjective taste for gears often enough over the years. But I wouldn't put him in a panel to discuss measurements or their interpretation.

Tool on the other hand, IDK if he's fun at parties, but he's been as solid as it gets on speaker research. So anytime their views diverge about measurements, you know where my vote goes. On more subjective stuff, it's another story.

Thank you for weighing in on this, castle.

Dr. Toole is da man when it comes to measurements, imho. I'm less familiar with Mr. Guttenberg. So not really sure what he's all about. Though it seems alot of audiophiles like the guy.

I told him already. It's written all over his posts that he's made for that stuff. And he often enough gets confused by a lack of personal experience with measurements. ”This is the way”.

We are all in different phases of learning about this stuff. Which is why forums like this exist! If there's a particular thing (or things) that you still feel that I'm confused on though, I'd be curious to know what that is. (?)

 

[ADUHF]

Haha let's make it happen then .
@ADUHF feel free to ask how to get started if I can help.

Thank you, Maya. And I'm going to hold you to this! If/when I decide to begin pursuing this, you will likely be one of the first to know.

 

[MayaTlab]

I'm not sure I'm following all of the above.

Since my probe is a very DIY thing, I dressed a list of "items of concerns" that I wanted to test for to increase my confidence in its results. Such as : is the sound truly coming from the orifice or leaking through the tube's walls or through the tube to mic attachment system ? For this one testing for it is fairly straightforward, just measure either my speakers or my headphones with the tube's orifice open vs. blocked.

But it dawned on me that since several articles seem to indicate that measurements done at the entrance of the ear canal should provide fairly similar relative results to measurements near the DRP (ie there could be the existence of a constant transfer function between the measurements done at the entrance of the canal and at the DRP, in other words the relative differences between headphones would be similar), if I were to get a constant transfer function between the two techniques' relative results it would turn most of my concerns into a moot point and avoid me the trouble of testing for them.
For example, I was wondering what the impact of tubes (either my probe's tube, or the mics' power cables) of various sizes, perhaps routed in different ways through the concha, would be. Or what would be the impact of the tube being compressed by the pads. But if no matter what I do in that regard I'm still getting a constant transfer function between the two, then I think it means that at least as far as relative comparisons are concerned it's a non-problem.

So that's what you're seeing in this graph : plotting the relative difference between my pair of HD650 (here with Dekoni Elite Velour pads, no not comparable to stock) vs three other headphones, and looking at the resulting curves for several measurement techniques.
If the traces align with each others, then it means there is a constant transfer function. If they don't, there isn't.
I'm both quite pleased and displeased by what I'm getting so far.
Quite pleased because up to 7kHz or so I'm getting curves that reasonably match each others, and in the region 2-4kHz where they don't to quite the same degree an explanation for the discrepancy can be found (blocked vs open ear canal). In both cases the degree they agree with each others is quite a bit tighter than the differences I'm getting when equalising HPs to the same target based on industry standard rigs / presets.
And displeased because in the 7-10kHz range I'm not getting a constant transfer function at all (although it's still fairly straightforward to know which one I should trust, but that involves listening tests).
It's entirely possible that my blocked ear canal measurements aren't done deep enough, so I'll try to move to another mic system, besides also moving to other microphones for the probe tube.

 

[ADUHF]

New livestream is up, with guest Crin. This is a pretty long one at 2:30. Later in the stream, Crin and Resolve get into a somewhat heated back and forth about whether or not both channels of a headphone's response should be shown on graph.

There are some good arguments on both sides of this question, but I tend to come down in favor of either showing both channels, or including a plot that shows just the relative differences between the two. Because most of the lower-end single-sided headphones I can afford will have some imbalances. And I like to see what some of the differences are, esp. if they're consistent across multiple units (which is somewhat hard to tell when looking at just one unit).

 

[ADUHF]

I've also started tinkering around a bit with some possible neutral reference curves for the 5128 HATS rig, based on the DF response of the rig and sound power curves of neutral loudspeakers.

The graph below is just the 5128 DF response curve combined with both a -1.0 dB per octave slope and a -1.5 dB per octave slope. Which is the approximate range of both the 2018 Harman over-ear headphone target with diffuse field compensation, and also the sound power response of neutral loudspeakers, on average.

The above curve is scaled to the same ratio as Jude's frequency response curves in the Head-Fi Audio Measurements Lab forum, to make comparisons a little easier. Below are the 5128 measurements of the Apple APM, Senn HD 650, and Senn 800s though for comparison to the above.

 

[Resolve]

And I like to see what some of the differences are, esp. if they're consistent across multiple units (which is somewhat hard to tell when looking at just one unit).

This is important, but I think the only way to accurately represent that is by having multiple units (and a lot of them). Then you'd be able to have a spread and show tolerances, which is interesting information and a demonstration of the company's QC. My main reasoning for not publishing that information is because with literally every brand, there's always a chance you could get unlucky, and unless you have multiple units you don't know which is representative. So for example, if you say "this happened to me, and therefore it might happen to you", that could be the case for quite literally any product/brand, and in the cases where you do get perfect channel matching, it would suggest that those brands are immune to this QC issue, which they're not necessarily. You just didn't happen to get unlucky with those ones.

And in my case, if there's audible channel imbalance, I'd simply send it right back and ask for a replacement, which typically manufacturers are willing to oblige. In fact, I'd go as far as to say that any representation of significant channel imbalance is more likely to not be representative, because they're generally within a certain tolerance margin. So it's more of a question about what's prudent to represent, and what readers would potentially take away from it. To be clear, I do measure both channels, and if there isn't good enough channel matching, I'll just try to get a different unit - maybe it's something I'll note in the review as well. But measurements have a tendency to persist, almost like "if you're buying headphone X, you're going to get THIS". And that's really where my worry comes from.

I know Crin disagrees of course and I'm totally sympathetic to his reasoning for showing it, and it was a fun discussion, but I think if the information I was aiming to publish was about the company's QC tolerances, I'd look to get a whole bunch of units in and track the channel balance delta over a series of units. And personally, I'm much more interested in ensuring it's a representative graph for how the headphone should measure - because in most cases you're more likely to get a unit that's within the margins than one that's outside of it... that's why those tolerances exist in the first place.

 

[ADUHF]

Welcome aboard, and thank you for weighing in and explaining your take a little more on the above, Resolve.

I always enjoy listening to these live streams btw, and frequently take new ideas away from them,... whether you have on any guests or not. I'm sorry for the bad heatwave you guys have been dealin with as well. I've spent most of my adult years in warmer climes, so I'm somewhat used to it by now. But I know for folks who aren't, that it can be a difficult adjustment.

I understand that the subject of driver symmetry is a thorny one, which can present some special challenges to reviewers like yourself. For both the reasons you've stated above. And for other potential reasons as well. For starters, everything has to be working pretty much perfectly on your equipment just to ensure that both earpieces on the headphone are receiving the exact same signal amplitude at all frequencies. I would imagine that there are some ways of calibrating the rig to improve or refine its performance along these lines though.

There could also be issues related to the way a headphone interfaces or seals with a rig, that could potentially introduce measurement artifacts into one channel, but not the other. Averaging the response over multiple seatings (as you presently do) should help to mitigate the chances of that occurring though, I would think.

There are also alot of people who simply aren't that sensitive to, or interested in L/R imbalances as well. And mostly concerned with a headphone's tonal balance. I.e. the levels in the bass versus the levels in the treble or midrange, for example. Rather than the levels of the right vs. the left channels. This is probably more the case at the higher end of the price scale than at the lower end, I suspect.

There could also be some imbalances at certain frequencies in your own hearing, that might make you less sensitive to some kinds of imbalances in a headphone's response than others. And less than an ideal subjective judge of problems along these lines. I have more HF loss in my right ear, for example, than in my left. Which sometimes makes it a bit more difficult for me to judge when a headphone is favoring the right channel a bit more. So humans are not always the best gauges for these kinds of things.

And there are probably some other issues along a similar line that I haven't thought of, which could make it more difficult to accurately assess or measure a headphone's L/R balance.

Most of the headphones that I've used have audible imbalances though, of one kind or another. And it's often helpful to see where and how that's occurring a little better on some type of a graph, both to confirm or deny what I may be hearing (or think I'm hearing) in the headphones with my own ears. And to also correct or compensate for it, if necessary (and possible).

At the higher end of the price scale, these imbalances may be the result of quality control issues, as you suggest. So I understand where you're coming from when you say that you'd generally just return a pair of headphones for repair or replacement, when you encounter something like that in most of the headphones that you're reviewing... That makes perfect sense.

At the lower end of the price scale, it can be a (potentially correctable) flaw in the actual design of the headphones though. And not something which varies significantly from unit to unit. This seems to be a very common issue, for example, with alot of headphones that have single-sided cables that attach to only one earpiece. Which is more common among the lower-cost studio type headphones that I mostly use than in higher-end headphones (which typically have double-sided cables).

All of the halfway decent headphones that I've personally owned have been single-sided. This includes the AKG K553 Pro, AT M50x and Beyer DT-770. And they have all had noticeable imbalances in their drivers, of one kind or another. Which I believe are less the result of QC issues. And likely more related to their (single-sided) designs. The same has been true of most other headphones that I've also tried or used (but decided not to buy or keep) in the same or similar price ranges as the above units. I could rattle off a fairly long list of them to you, but it would include other popular lower-end headphones like the AKG K371, and AT M40x. And possibly also one or two of the Sennheisers.

Single-sided cables have unfortunately become kind of a de facto standard, especially for studio headphones. But also for alot of consumer headphones in the sub-$300 range. So they are difficult to avoid at this point. And it's something I've frankly never really liked that much or appreciated, even for my own "studio-ish" work. And if you decide to exclude all headphones with this type of design, it extremely limits the available options.

If you know what the imbalances are, then in many cases they can be corrected. Which can sometimes result in a dramatic improvement in the headphone's stereo imaging performance! If there is just a slight difference in the overall volume between the two drivers, like on my 250-ohm DT-770's, then that's a fairly easy issue to fix. Because it can be done with just a simple left-right stereo balance control (if you happen to have one of those on your player, amp, or DAC... which many people do not!).

If the drivers have different imbalances at different frequencies though, then correcting the issue is more difficult, because it may require separate EQs, or tonal balance adjustments for the two channels at different frequencies, which is more involved. If the headphones are easily reversible (like my previous AKG K553, and the Beyer DT-770), then there are some tests you can do with frequency sweeps and whatnot, to figure out a little better where the L/R discrepancies are in the frequency range, to better correct them. Imbalances in your own hearing also have to be taken into account when doing this though. And in-ear microphones could maybe also be helpful in looking at this.

If you can see the same problem in multiple reviews, with multiple units on different graphs though, then it's potentially a bit easier to localize and correct. The problem with this is that many reviewers receive the same review unit from a manufacturer. So you don't really get to see how this sort of problem can vary (or stay the same) from unit to unit.

This is a shortcoming in headphone reviews in general though, which potentially effects the reliability and accuracy of a reviewer's impressions and measurements across the board. And is not just related to the issue of a headphone's driver symmetry. So I don't really understand why you'd want to single this one particular aspect of a headphone's performance out for exclusion from a review, in this regard. And not some of the other aspects, such as the headphone's tonal balance, for example. Because there could potentially also be some differences in a headphone's sound and performance that varies from unit to unit in these other areas as well... No?

I agree with you that good driver symmetry should just be a "pass/fail" kind of thing. But I've found that's not quite the way things work in reality at the lower end of the price scale. And the only way to really change that circumstance is to educate people on the issue, through reviews (and discussions like this one), so that the manufacturers do a better job of addressing the issue in their products at all price points. This is my opinion anyway.

A longish way of just sayin I mostly agree with Crin's take on this particular issue. But there ya go. Hopefully some of this made a little sense.

 

[ADUHF]

Here is a graph from a review of my old AKG K553's that shows some of the imbalances in its drivers, which varied noticeably across the frequency range...

https://reviewed-production.s3.amazonaws.com/attachment/47774252831e427a/AKG_K553_Tracking_Chart.jpg

This is the kind of imbalance that is more difficult to correct. And since I was only able to find one review that showed this type of info for a single unit, I couldn't really use the above plot to reliably correct the problem on my own pair of K553's.

There were other potential QC and fit/seal-related issues that could also have contributed to the above results, as well as the headphone's single-sided cable design. So for a variety of reasons, I had to try to figure out my own solutions to correcting the problem (as best I could) on my particular unit, without the aid of graphs like the one above.

The headphone's stereo imaging improved noticeably imo after making some adjustments to fix its imbalances at the different frequencies though. And more reviews of different units, with more plots like the one above might possibly have helped somewhat in that process.

The original AKG K553 Pro is probably one of the more difficult headphones to measure accurately though, imo. Because of its large, and low-hanging (and often poor-sealing) earpads. And because there are some poorly-attached filters/dampers on its drivers, which can easily come loose and change its sound on one side or the other. Hopefully the issues with the filters have been corrected on the current MKII version with the detachable cable.

These are the sort of things that I would hope to see discussed in a really good review of the headphones though.