ASR Headphone Testing and BK 5128 Hats Measurement System

[Robbo99999]

Some good questions here.

In broad terms, I think the premise that high-frequency equalization should mostly take the form of shelf or very low Q peak filters (broad adjustments in level, essentially) is quite advisable. I personally take this to a relative extreme; I don't even aim to notch the peak from my HD800, because I've found that doing so without putting a hole in the surrounding treble somewhere is troublesome. For most practical people's purposes, so long as it doesn't sound bad, I'd say that relatively fine/high Q equalization is reasonable up to 8-10khz, although it depends on a number of variables (particularly how the headphone's response varies with position on the head of the wearer).

Regarding the intersection of HRTF and HpTF, let's consider two extremes and how they would impact things:

First, let's imagine a world where headphones are "HRTF chameleons" - by some process, whether acoustic or computer controlled, they perfectly approximate the individual wearer's HRTF in the target sound field. In this world, you could see extremely wide variation in HRTFs and HpTFs, but have zero variation in subjective timbre of headphones, because headphone subjective frequency response is equal to HpTF minus HRTF. This world is almost entirely reconcilable with Hammershøi & Møller's data.

Second, let's imagine a world where headphones are "HRTF blind" - perhaps in this world a trend of very deeply inserted in-ear monitors dominates, but for whatever reason, the HpTF is absolutely constant between wearers, even as individual HRTF varies. Subjective frequency response would track roughly according to the scenario you're outlining in this post - up to around the peak of the ear resonance there'd be agreement, and then it would all go south. A headphone that sounds peaky to one person would be smooth to another, and we'd have a great deal of trouble comparing our subjective impressions of headphones at all in the higher frequencies.

Now, in reality, we don't live in either of those worlds - some of the aspects of individual anatomy that influence HRTF also influence HpTF, so they aren't uncorrelated, but they also don't coincide perfectly. There's also the "x factor" of whether headphones have individualized and atypical interactions with some anatomy that doesn't relate in any way to HRTF - unarguably this happens at low frequencies with headphones with high acoustic impedance when a leak is present in the pad volume, but it could also happen at higher frequencies, and this would introduce another source of response variation.

Pragmatically, I don't think that we can use measurements of headphones on population average measurement fixtures to make some of the projections we can make about other types of equipment - e.g. "DUT A will sound the same as DUT B" - and non-individualized equalization is inevitably going to be an area where caution is wise, and erring towards broader filters and leaving the high Q features mostly alone is likely to yield better results on average, but equally, headphones are such radically audibly different devices that we don't need the degree of consistency of coherency between test and in situ that we have with amplifier or DAC measurements to make very reasonable extrapolations about what will sound better.

This all said, I still don't entirely grasp what Amir's primary goal with this headphone testing project is; if it's to redefine headphone metrology and take it out of the dark ages...well, we weren't in the dark ages to begin with, so that'd be pretty hard If it's to present an additional impartial source of reliable headphone measurements alongside what presently exists (Oratory, Resolve/Headphones.com, Clarityfidelity/Speakerphone, Keith Howard/HeadphoneTestLab, Brent Butterworth/Soundstage Solo, etc), then we've already got validation of concept. If it's to improve on the current state of the art in headphone metrology that's very conceivable - the 5128 can reasonably claim to be the most accurate way to measure headphones that presently exists, and there's still plenty of room for innovation in methodology if Amir is interested in that.

Thanks for that response, I had some time finally today to digest what you were saying in this post I think I understand all of it apart from the bit of your post that I put in large font & underlined: "because headphone subjective frequency response is equal to HpTF minus HRTF". I can't really visualise that and I don't know exactly what you mean by subtracting one HRTF from another, are you talking about literally subtracting the two frequency response curves from one another, and still I can't visualise the applicability meaning of that? But I do understand & can visualise the paragraph that follows that one & the rest of your post.

 

[Mad_Economist]

Thanks for that response, I had some time finally today to digest what you were saying in this post I think I understand all of it apart from the bit of your post that I put in large font & underlined: "because headphone subjective frequency response is equal to HpTF minus HRTF". I can't really visualise that and I don't know exactly what you mean by subtracting one HRTF from another, are you talking about literally subtracting the two frequency response curves from one another, and still I can't visualise the applicability meaning of that? But I do understand & can visualise the paragraph that follows that one & the rest of your post.

I'll see if I can get some visual aids and/or data together later today - although this may need to wait for a later time, I'm a tad busy at present - but think of it this way: Imagine that the diffuse field HRTF is the right one for subjectively flat frequency response. In this case, we'd measure the DF-HRTF of a head, then measure the response of a headphone on the head, and subtract the DF-HRTF from the response we measured for the headphone - the resulting frequency response plot will match the subjective perception of the headphone. For a perfect match - absolute subjective neutrality - it'd be a flat line.

 

[Robbo99999]

I'll see if I can get some visual aids and/or data together later today - although this may need to wait for a later time, I'm a tad busy at present - but think of it this way: Imagine that the diffuse field HRTF is the right one for subjectively flat frequency response. In this case, we'd measure the DF-HRTF of a head, then measure the response of a headphone on the head, and subtract the DF-HRTF from the response we measured for the headphone - the resulting frequency response plot will match the subjective perception of the headphone. For a perfect match - absolute subjective neutrality - it'd be a flat line.

Ah, when you talk about "measure the response of a headphone on the head, and subtract the DF-HRTF from the response we measured for the headphone" and that a perfect match would be a straight line......you're referring to the Compensated Graph or Compensated Curve that I briefly referred to in this post for instance (https://www.audiosciencereview.com/forum/index.php?threads/headphone-measurements-using-brüel-kjær-5128-hats.15352/post-486426)? In that case I understand what you mean, I've never been a fan of those compensated graphs, to me they're not intuitive and don't aid my visualisation of what is occurring.

Ah, and when you talk about "subjective frequency response" you're referring to the fact that what we hear as flat is not actually a flat Frequency Response (visually on the graph) unless we're talking about the Compensated Graph/Curve, in which case the ideal is a flat curve. Right, I think I understand you now. It's the old classic Compensated Graph confusing me and not meshing easily with my brain! I hate thinking/visualising stuff based on using the Compensated Graph.

 

[Mad_Economist]

Ah, when you talk about "measure the response of a headphone on the head, and subtract the DF-HRTF from the response we measured for the headphone" and that a perfect match would be a straight line......you're referring to the Compensated Graph or Compensated Curve that I briefly referred to in this post for instance (https://www.audiosciencereview.com/forum/index.php?threads/headphone-measurements-using-brüel-kjær-5128-hats.15352/post-486426)? In that case I understand what you mean, I've never been a fan of those compensated graphs, to me they're not intuitive and don't aid my visualisation of what is occurring.

Ah, and when you talk about "subjective frequency response" you're referring to the fact that what we hear as flat is not actually a flat Frequency Response (visually on the graph) unless we're talking about the Compensated Graph/Curve, in which case the ideal is a flat curve. Right, I think I understand you now. It's the old classic Compensated Graph confusing me and not meshing easily with my brain! I hate thinking/visualising stuff based on using the Compensated Graph.

Indeed - and this is the strong argument precisely for compensation. With the proper response target subtracted, you are actually seeing a plot of the subjectively perceived timbre/frequency response of the headphone. That is, a flat line sounds neutral, a line slanting down to the right sounds bass-heavy or treble-lite, and vice-versa for a line slanting upwards to the right.

 

[Robbo99999]

Indeed - and this is the strong argument precisely for compensation. With the proper response target subtracted, you are actually seeing a plot of the subjectively perceived timbre/frequency response of the headphone. That is, a flat line sounds neutral, a line slanting down to the right sounds bass-heavy or treble-lite, and vice-versa for a line slanting upwards to the right.

Yep, I see that, I prefer to use the Raw Frequency Response Graphs overlaid on the Frequency Response Target of the Harman Curve, and my mind can see the deviation above & below from that, so I see the compensated graph as an unnecessary complication, but I do know what you mean. Thanks for explaining!

 

[March Audio]

What I want to do is data dependent. There are two questions here:

1. Do we get into headphone testing at all?

2. Do we do it with 5128?

To go with 5128, it needs to show substantial value over existing rigs. If the difference is very minor, then it makes no sense to pay its huge premium.

We have to somehow quantify 5128's value. In a sea of measurement variability this is challenging but we need to do that in the next few days before I have to return the thing.

I have been looking through the other thread of measurements.

The thing that concerns me is that considering how every individuals ear canal and pinna affect how they hear any of these headphones, how much value will we get from the measurements? I suppose we can say whether a headphone conforms to the recommended curves, but beyond that? As such I struggle a bit with the significant outlay. Would it be better to have the lower cost system and spend the saved money on something else?

Not sure what, maybe a part time assistant to take some of the testing workload off you ?

 

[Mad_Economist]

The thing that concerns me is that considering how every individuals ear canal and pinna affect how they hear any of these headphones, how much value will we get from the measurements?

This isn't just pointed at you - quite a lot of people are saying roughly this - but since you're here, what leads you to believe that this effect is more significant with headphones versus speakers, as an example? If anything, a speaker in a room - with a large share of reflected sound at many frequencies in the HRTF band - is surely a case where individual anatomic variation would present itself quite strongly, no? And yet, we see pretty strong groupings in preference with speakers based on anechoic measurements with freefield mics; and the same with headphones with standardized test fixtures.

This said,

Would it be better to have the lower cost system and spend the saved money on something else?

this is entirely conceivable - I would be lying if I said that I thought the 5128 is a good value when it's a multiple of other options...but it is cool, and speaking selfishly, the headphone community would benefit by the opportunity to see more data from a 5128, as we currently have no sites publishing with one.

 

[March Audio]

This isn't just pointed at you - quite a lot of people are saying roughly this - but since you're here, what leads you to believe that this effect is more significant with headphones versus speakers, as an example? If anything, a speaker in a room - with a large share of reflected sound at many frequencies in the HRTF band - is surely a case where individual anatomic variation would present itself quite strongly, no? And yet, we see pretty strong groupings in preference with speakers based on anechoic measurements with freefield mics; and the same with headphones with standardized test fixtures.

This said,

this is entirely conceivable - I would be lying if I said that I thought the 5128 is a good value when it's a multiple of other options...but it is cool, and speaking selfishly, the headphone community would benefit by the opportunity to see more data from a 5128, as we currently have no sites publishing with one.

The coupling mechanism to your ear. The size and shape changes the resonances. Its an enclosed chamber (lossy). Its not the same as listening to "the room". The variation is increased.

No we dont see strong groupings in speaker preferences. We see that people prefer a flat on axis anechoic response and a smooth off axis response. So if listening to headphones was the same all we would need is a flat response measured in free space. Doesnt seem to work like that does it ?

Agreed we see correlation to the recommended curves but do you need the B&K to achieve the necessary level of accuracy to judge conformance?

 

[Mad_Economist]

The coupling mechanism to your ear. The size and shape changes the resonances. Its an enclosed chamber (lossy) Its not the same as listening to "the room".

It is indeed not the same as listening to the room, and you're quite correct that the eardrum sound power will vary for the same headphone on different individuals; my point is that this is true also of speakers and rooms, but that with both speakers and headphones, we can make robust inferences regarding listener preference without individualized measurements, based on measurements in the speaker case in anechoic conditions, and in the headphone case on generic test fixtures.

No we dont see strong groupings in speaker preferences. We see that people prefer a flat on axis anechoic response and a smooth off axis response.

This steps slightly outside of my silo in headphone world, but while it's been a while since I read Olive 2004, my rough read of Toole and Olive is that the following statements are true:

1: People like speakers that do not have substantial frequency response differences between their on- and off-axis radiation - that is, the timbre isn't different for the reflected sound.
2: People like speakers whose in-room power response tends to be downward sloping.
3: Anechoically flat on-axis speakers with limited higher-frequency radiation patterns but well controlled directivity meet both of the above requirements.

It certainly seems accurate from that to say that a flat anechoic response and smooth off-axis response will be liked, but I'm not sure it's necessary that a speaker be anechoically flat and be directive at higher frequencies to achieve this result; I recall that some non-flat-anechoic omnidirectional speakers have scored acceptably in listening tests, for example. This is a pretty minor quibble however, I think that we agree on how speaker preference correlates to measurements.

Edit: My apologies, I started responding before your edits, I'll rectify that in a separate comment to avoid circular "oops I missed the edits".

 

[Blumlein 88]

This isn't just pointed at you - quite a lot of people are saying roughly this - but since you're here, what leads you to believe that this effect is more significant with headphones versus speakers, as an example? If anything, a speaker in a room - with a large share of reflected sound at many frequencies in the HRTF band - is surely a case where individual anatomic variation would present itself quite strongly, no? And yet, we see pretty strong groupings in preference with speakers based on anechoic measurements with freefield mics; and the same with headphones with standardized test fixtures.

This said,

this is entirely conceivable - I would be lying if I said that I thought the 5128 is a good value when it's a multiple of other options...but it is cool, and speaking selfishly, the headphone community would benefit by the opportunity to see more data from a 5128, as we currently have no sites publishing with one.

Point it at me if you wish. I think the same thing. You've got an expensive measurement system with lots of fine tuning finagling every step of the way and still the end result if very, very variable. It is no where close to as useful or consistent as the speaker measurements. Looks like it never will be with this approach.

My brain calibrates my ears for my HRTF and in a room with speakers we all have more or less equal perceptible parameters. Not so with headphones that get too close to and distort the effects of the HRTF for each invididual. That brain calibration doesn't help when you stick a close up headphone driver right on your ear and disrupt that natural process. It is somewhat failed from the start.

 

[Mad_Economist]

So if listening to headphones was the same all we would need is a flat response measured in free space. Doesnt seem to work like that does it ?

Relatively speaking, it's fairly close. Consider Olive, Welti, and Khonsaripour 2018 - adherence to the Harman target frequency response on a generic fixture is the only actual measurement input, and the prediction is fairly robust.

Agreed we see correlation to the recommended curves but do you need the B&K to achieve the necessary level of accuracy to judge conformance?

Absolutely not, and tell anyone who says so to wash out his mouth with soap, as he's a liar - the 5128 is cool, but there's a reason I generally advise folks to get 43AGs.

 

[March Audio]

It is indeed not the same as listening to the room, and you're quite correct that the eardrum sound power will vary for the same headphone on different individuals; my point is that this is true also of speakers and rooms, but that with both speakers and headphones, we can make robust inferences regarding listener preference without individualized measurements, based on measurements in the speaker case in anechoic conditions, and in the headphone case on generic test fixtures.


This steps slightly outside of my silo in headphone world, but while it's been a while since I read Olive 2004, my rough read of Toole and Olive is that the following statements are true:

1: People like speakers that do not have substantial frequency response differences between their on- and off-axis radiation - that is, the timbre isn't different for the reflected sound.
2: People like speakers whose in-room power response tends to be downward sloping.
3: Anechoically flat on-axis speakers with limited higher-frequency radiation patterns but well controlled directivity meet both of the above requirements.

It certainly seems accurate from that to say that a flat anechoic response and smooth off-axis response will be liked, but I'm not sure it's necessary that a speaker be anechoically flat and be directive at higher frequencies to achieve this result; I recall that some non-flat-anechoic omnidirectional speakers have scored acceptably in listening tests, for example. This is a pretty minor quibble however, I think that we agree on how speaker preference correlates to measurements.

Edit: My apologies, I started responding before your edits, I'll rectify that in a separate comment to avoid circular "oops I missed the edits".

Well maybe this is anecdotal, but FWIW when I look at discussions of headphones I see far wider ranging views regarding the perceived sound. My own experience indicate how much I can literally detest some headphones (often highly rated by others) in a way I never do with speakers . So Im not convinced we are making such robust inferences as we do with speakers.

 

[March Audio]

Relatively speaking, it's fairly close. Consider Olive, Welti, and Khonsaripour 2018 - adherence to the Harman target frequency response on a generic fixture is the only actual measurement input, and the prediction is fairly robust.View attachment 78572


Absolutely not, and tell anyone who says so to wash out his mouth with soap, as he's a liar - the 5128 is cool, but there's a reason I generally advise folks to get 43AGs.

Is a "generic fixture" free space? The point of my comment was comparing to the response of speakers and the perception there of.

If it were the same we would just need a flat free space response, probably quite easy to achieve with a small headphone driver, and that would be that.

 

[Mad_Economist]

Point it at me if you wish. I think the same thing. You've got an expensive measurement system with lots of fine tuning finagling every step of the way and still the end result if very, very variable. It is no where close to as useful or consistent as the speaker measurements. Looks like it never will be with this approach.

As you wish.

You're correct that the results are highly variable in situ, but the same can of course be said of an unwindowed in-room measurement of a speaker at an arbitrary position. The different with speakers is that we've got a fairly nifty toolkit for divorcing ourselves from the reality of the playback situation for practical measurements, whereas with headphones, you must take the whole kit and caboodle as well. This reflects the reality of headphones on heads, however - the audible effects of positional variation in headphones are meaningful, just as moving around within a listening room is meaningful to our subjective perceptions.

My brain calibrates my ears for my HRTF and in a room with speakers we all have more or less equal perceptible parameters. Not so with headphones that get too close to and distort the effects of the HRTF for each invididual. That brain calibration doesn't help when you stick a close up headphone driver right on your ear and disrupt that natural process. It is somewhat failed from the start.

This is where you're going off the reservation. The inverse HRTF filtering process we apply to speakers we apply as well to headphones (albeit we don't use the inverse filter set for a frontal sound source; see Theile 1986, 2016, etc), and the same rules for timbral accuracy apply: we need eardrum sound power to approximate the listener's head in the sound field of the perceived acoustic source (arguably a diffuse field, see Theile if you care but honestly nobody does these days) to perceive the sound as "correct", or "flat", or what have you.

In a scenario where HRTF varied wildly - which, to some degree, it does, at least at some frequencies - and the response at the eardrum of headphones was absolutely constant across users, you would be quite correct that we would be unable to predict subjective tonality above perhaps 2khz or so. In such a world, each headphone would effectively be different to each user because it was the same and the users differed. I've spent a fair amount of time measuring headphones on different pinnae, however, and I simply do not see this effect - there are meaningful differences between the same headphone measured on two different sets of ears, and two different headphones on two sets of ears tend to differ in similar ways, at least for the circumaural designs that make up the majority of the high-end market. I'll freely acknowledge this as a caveat of any design that bypasses the ear anatomy - and IMO it contributes to the higher variation of preference in in-ear designs, which bypass the ear almost entirely - but it's my opinion that you just can't reconcile the degree of skepticism you're voicing with the degree of consistency in ex. Sean Olive's correlation of headphone response on a mannequin to subjective assessments.

 

[Mad_Economist]

Well maybe this is anecdotal, but FWIW when I look at discussions of headphones I see far wider ranging views regarding the perceived sound. My own experience indicate how much I can literally detest some headphone in a way I never do with speakers . So Im not convinced we are making so robust inferences as we do with speakers.

I would make the metaphoric comparison to the days of speakers before the work of Floyd Toole, in terms of headphones - at present, you fairly rarely encounter a speaker whose frequency response is grossly deviating from anechoically flat, or whose directivity is completely random. This wasn't always the case, and in the darker ages of speaker design, there was greater variation in preference partially because people were choosing their relative poisons. I would argue that this is roughly where we are today in headphones.

Is a "generic fixture" free space? The point of my comment was comparing to the response of speakers and the perception there of.

If it were the same we would just need a flat free space response, probably quite easy to achieve with a small headphone driver, and that would be that.

I believe that one of us is missing the other's point - I am not and have not been arguing that headphones approximate speakers in terms of listening conditions. My question initially was why you would assume greater subjective variation in the headphone case specifically.

It would definitely be extremely incorrect to say that - outside of the case of continuous head-tracking and HRTF emulation - headphones approximate the experience of listening to speakers at either a subjective level or in terms of what's going on at any given point in our aural anatomy.

 

[Blumlein 88]

snippage...........

In a scenario where HRTF varied wildly - which, to some degree, it does, at least at some frequencies - and the response at the eardrum of headphones was absolutely constant across users, you would be quite correct that we would be unable to predict subjective tonality above perhaps 2khz or so. In such a world, each headphone would effectively be different to each user because it was the same and the users differed. I've spent a fair amount of time measuring headphones on different pinnae, however, and I simply do not see this effect - there are meaningful differences between the same headphone measured on two different sets of ears, and two different headphones on two sets of ears tend to differ in similar ways, at least for the circumaural designs that make up the majority of the high-end market. I'll freely acknowledge this as a caveat of any design that bypasses the ear anatomy - and IMO it contributes to the higher variation of preference in in-ear designs, which bypass the ear almost entirely - but it's my opinion that you just can't reconcile the degree of skepticism you're voicing with the degree of consistency in ex. Sean Olive's correlation of headphone response on a mannequin to subjective assessments.

I seem to experience what you say you aren't finding. Maybe my HRTF is an outlier. Binaural doesn't much work. Phones are definitely a borrow or buy and try situation. Highly recommended ones sound yuck to me, others sound okay, none sound good.

The most subjective cursory review of speakers seems to have some merit most of the time, while the same of headphones is simply rolling the dice for me. I for a long time used some Nakamichi SP7 on ear phones which were okay. I've had Stax Lamda, Omega, Koss ESP9, DT880, various AKG, Senn's which I've never found very useful, and Sony 7510 which is the best I've used. Which isn't to say they are great they are just not bad vs other phones. Maybe okay 80 hz to 5 khz for me.

I understand the approach being taken here. It all makes good sense right up until everyone papers over how different HRTF's and pinna are among people, and also average or ignore how radically different measured results above 5 khz are with simply removing phones and replacing them. Our ears have evolved with mechanisms to ignore the space within which a sound occurs which allows us to ignore the space largely within which speakers function. The same cannot be said for sticking a sound transducer over our ears and playing sounds.

 

[March Audio]

I would make the metaphoric comparison to the days of speakers before the work of Floyd Toole, in terms of headphones - at present, you fairly rarely encounter a speaker whose frequency response is grossly deviating from anechoically flat, or whose directivity is completely random. This wasn't always the case, and in the darker ages of speaker design, there was greater variation in preference partially because people were choosing their relative poisons. I would argue that this is roughly where we are today in headphones.


I believe that one of us is missing the other's point - I am not and have not been arguing that headphones approximate speakers in terms of listening conditions. My question initially was why you would assume greater subjective variation in the headphone case specifically.

It would definitely be extremely incorrect to say that - outside of the case of continuous head-tracking and HRTF emulation - headphones approximate the experience of listening to speakers at either a subjective level or in terms of what's going on at any given point in our aural anatomy.

Im not so sure that is that is the case. "Choosing your poison" is still very pertinent today and as we have seen from Amirs tests speakers still do vary quite considerably in their responses.

Aaahhh. Im not presuming subjective variation, Im referring to the physiological variation.

 

[Mad_Economist]

If you'll allow me, for a moment, I think that your misconception here is that your experience does not align with what I am saying rather than anything else.

I seem to experience what you say you aren't finding. Maybe my HRTF is an outlier. Binaural doesn't much work

Binaural recordings are individualized, and since individual HRTF is indeed quite variable, they are the closest thing you can find to "putting on a pair of someone else's ears". This is their failing, as you've correctly identified - the difference between your HRTF and the recording head's HRTF determines the functionality of binaural recordings for you; generic recordings very rarely work for me either.

Phones are definitely a borrow or buy and try situation. Highly recommended ones sound yuck to me, others sound okay, none sound good.

I'm going to point very slightly upthread to reference myself here, but I really feel the need to emphasize this: by the standards of Sean Olive's work, a very large share of headphones cannot be expected to perform well in subjective listening tests, including a lot of expensive models. Why people like these headphones, in many cases, is a much more complicated question - if you feel cynical, you could point out that some fundamentally flawed breeds of speakers (fullrange single-driver boxes, anyone?) also have their own committed fandoms, and of course sound isn't necessarily the only game in town for preference - but that's all kind of aside to the point that none of this variation itself is contraindicated by what we know about the correlation of measured frequency response and subjective preferences.

none sound good.

Honing in on this for a moment, while I'm reaching here, this suggests to me that you haven't run into a headphone that matches the qualities - and quality, in your evaluation - of the speaker listening experience. This is entirely in line with what is to be expected per the normal body of the lit: since the 80s at least, it's been known that you cannot replicate the experience of subjective spatiality without binaural sound cues, and that headphone sound is localized within the head, with negative impacts both in subjective tone and the experience of music in general. Particularly key for your potential listening pleasure, there's been some work done over this time to try to address this. A recent paper that might be interesting to you is Theile 2016, because the failings of headphones to replicate the experience of speaker and live listening are precisely Theile's concern. He also references some recent commercial efforts to improve on the subjective experience of headphone listening - the Smyth Realizer, which if you feel like setting a titanic sum of money on fire for an experiment really would be worth a shot if your complaint about headphones is "sounds worse than speakers".

There's also some less, er, theoretical-and/or-wallet-annihilating work in this area. Head tracking has been shown to improve out of head localization even without individualized HRTF compensation (I can cite this if you'd like, but I'm like 85% sure Theile already does somewhere in my link), more and more databases of HRTFs have been made available for use with utilities like HeSuVi, and you even have cool projects like JaakkoPasanen's Impulcifier, which specifically sets out to allow you to individually compensate your own headphones.

This longwindedness to say that I think that your misconception is that the Olive research - and the "analysis from linear behavior on a generic mannequin" paradigm in general - is making a claim to something it isn't. I wouldn't expect to be able to say "this headphone will be a comparable experience to your speaker system based on its match to a frequency response target on my dummy" - but I can definitely say "headphone A is very likely to be preferred to headphone B given their measurements"...and there are other, separate research programs in improving headphone playback fidelity.

Anyway, sorry for stuffing words in your mouth on this one, but I thought I'd take a stab at bridging the apparent gulf of perspective.

 

[March Audio]

I seem to experience what you say you aren't finding. Maybe my HRTF is an outlier. Binaural doesn't much work. Phones are definitely a borrow or buy and try situation. Highly recommended ones sound yuck to me, others sound okay, none sound good.

The most subjective cursory review of speakers seems to have some merit most of the time, while the same of headphones is simply rolling the dice for me. I for a long time used some Nakamichi SP7 on ear phones which were okay. I've had Stax Lamda, Omega, Koss ESP9, DT880, various AKG, Senn's which I've never found very useful, and Sony 7510 which is the best I've used. Which isn't to say they are great they are just not bad vs other phones. Maybe okay 80 hz to 5 khz for me.

I understand the approach being taken here. It all makes good sense right up until everyone papers over how different HRTF's and pinna are among people, and also average or ignore how radically different measured results above 5 khz are with simply removing phones and replacing them. Our ears have evolved with mechanisms to ignore the space within which a sound occurs which allows us to ignore the space largely within which speakers function. The same cannot be said for sticking a sound transducer over our ears and playing sounds.

We are thinking very much along the same lines and have very similar experience - except that we wouldnt agree on which can sounded right . However I would wager we have similar thoughts about which speaker sounded right.

 

[Mad_Economist]

Im not so sure that is that is the case. "Choosing your poison" is still very pertinent today and as we have seen from Amirs tests speakers still do vary quite considerably in their responses.

Perhaps it is the dosage that's differing, then Put it this way: at present, I would not necessarily expect a headphone's response to even strongly resemble an HRTF when it's on a head, even if it's a highly priced model. It's comparatively rare to see a speaker that's even close to as far away from the anechoic targets as almost all headphones are from the on-HATS response targets we have.

Aaahhh. Im not presuming subjective variation, Im referring to the physiological variation.

Ah, yes, this clarifies it - physiological variation will absolutely occur, but its impact on subjective response depends on how correlated HpTF (headphone eardrum response) and HRTF-under-target-conditions (I didn't used to have to use these annoying constructions when DF-HRTF was the only target...) differ on an individual basis.