• 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!

Master Complaint Thread About Headphone Measurements

If the discussion was about frequency responce I would agree on this point. I not sure about distortion products and how it pertains to individual ear canal variation. It seems like a narrow topic, although I prefer lower THD whereas I appreciate the value of subjective evaluation of FR and distortion given the numerous unknown factors of headphone measurements.
thd and ear channel variance are two different things, but to be honest i visit ASR for the Klippel measurements, to me headphones measurements are kind of worthless, use the headphone... way more usefull than a graph of FR, in headphone-world. Even with head-fi tier of measurements are you ok, you only need the THD for know 2 things:
peak of distortion or too much distortion in it's sub-bass, the rest is pure waste of time, FR are NOT going to be nearly as useful as wear the headphone.
 
thd and ear channel variance are two different things, but to be honest i visit ASR for the Klippel measurements, to me headphones measurements are kind of worthless, use the headphone... way more usefull than a graph of FR, in headphone-world. Even with head-fi tier of measurements are you ok, you only need the THD for know 2 things:
peak of distortion or too much distortion in it's sub-bass, the rest is pure waste of time, FR are NOT going to be nearly as useful as wear the headphone.
Would you demo this headphone?

1718208861553.png



I personally wouldn't after seeing this FR. OTOH I'm less inclined to draw conclusions based on more granular features on these graphs.
 
thd and ear channel variance are two different things, but to be honest i visit ASR for the Klippel measurements, to me headphones measurements are kind of worthless, use the headphone... way more usefull than a graph of FR, in headphone-world. Even with head-fi tier of measurements are you ok, you only need the THD for know 2 things:
peak of distortion or too much distortion in it's sub-bass, the rest is pure waste of time, FR are NOT going to be nearly as useful as wear the headphone.
Not really, lots of people are going to enjoy an EQ to Harman Curve maybe with a personally tweaked bass level. In my experience & also from the Harman Research it's crazy to say that frequency response measurements are a waste of time. It's not as cut & dried as speaker measurements, but there is a lot of value in frequency response measurements for headphones - you just have to see the positive notes people write about Oratory's EQ's for instance, both here & on his reddit.
 
you just have to see the positive notes people write about Oratory's EQ's for instance, both here & on his reddit.
And what about negatives? the whole Oraty EQ present are another waste of time, everytime i recomended people pointed how bad it sounds like. Guys you can't measure headphones, measurements are worthless here and you can't use them to mark the quality from a headphone.
Your ear channel are going to transform the Harman curve into a Carlos curve anyway, and will be unpredictably
What do you mean by mask?
My point still stand.
 
My point still stand.
Statistically there exist trends between FR readings and subjective preference. I can only agree with the sentiment of your point. Measurements can be overvalued and undervalued youre taking an extreme stand on the later side but the data can't be ignored which point to mixed results more than useless results.
 
Statistically there exist trends between FR readings and subjective preference. I can only agree with the sentiment of your point. Measurements can be overvalued and undervalued youre taking an extreme stand on the later side but the data can't be ignored which point to mixed results more than useless results.
You didn't catch the point, there is no trend in nothing, measurements are worthless because your ear channel is unique, the '' ear '' from the measurement ring is not your ear, so.. it's worthless for you, is useful for someone who got the exact same ear as the '' ear simulator '' in the measurement rig.


headphones are kind of very very personal thing, sure you can tune to the harman curve the hadphone, but a '' defect '' in the harman curve can be a fixed by the ear channel
 
And what about negatives? the whole Oraty EQ present are another waste of time, everytime i recomended people pointed how bad it sounds like. Guys you can't measure headphones, measurements are worthless here and you can't use them to mark the quality from a headphone.
Your ear channel are going to transform the Harman curve into a Carlos curve anyway, and will be unpredictably

My point still stand.
Of course there are people that don't like Harman Curves or subtle variations of them - they're likely to be a minority rather than a majority. It doesn't matter too much that your ear canal, etc can change the frequency response you receive at your own eardrum, everyone's will be different to some extent, but the Harman Research has a large element of preference basing (& of course that will be amoungst people with different anatomy that changes the frequency response you receive at your eardrum). I mean even if people listen to Anechoic Flat Speakers in a room they will have different frequency responses at their eardrums, that doesn't make it invalid at all - in fact our brain normalises & gets used to it's unique signature that affects what you receive at your eardrum, it learns what sounds normal from everyday life. People still prefer Anechoic Flat Speakers in a room, it doesn't really matter that the frequency response they receive at their eardrum will be different for each participant (because that benchmark sounds natural to "everyone"), and likewise it doesn't matter that it'll be different received eardrum frequency response for each participant when wearing any given headphone. I suppose there can be some peculiarities of people's anatomy that will throw off some headphone models or make Harman Headphone Target sound weird/wrong, but either way you can't expect people to receive the same frequency response at each of their eardrum, as that's not part of the success of Anechoic Flat Speakers in a room nor indeed the Harman Headphone Curve.

EDIT: albeit the extrapolation that everyone receives a different frequency response at their eardrum when listening to ideal Anechoic Flat Speakers in a room, means that that there can't be one ideal frequency response created in a headphone that suits everybody to absolute perfection, instead to meet perfection would require a personalised Target Curve created by measuring Anechoic Flat Speakers in an ideal listening room with in ear mics, and then you match that target when wearing headphones - along the lines of the Smyth Realizer. So yes, Headphone Harman Curve is never gonna be 100% ideal, but then again we don't all have access to a Smyth Realizer and all the time to set it up & access to good listening rooms within which to do the measurements, etc. Harman Headphone Curve is a very good approximation for a lot of people until that point, and if you really don't like it then you just have to experiment to find out what is your ideal approximated curve, so then you can EQ your headphones to that new target, but it's not easy to do....for most people Harman Headphone Curve works or subtle variations thereof. To say that measured frequency response in headphones doesn't matter is ridiculous.
 
Last edited:
Your ear channel are going to transform the Harman curve into a Carlos curve anyway, and will be unpredictably
While ear impedance, pinna format and canal modes are 100% a real thing, the Harman curve is not the curve of a guy named Harman, rather an average of what a substantial amount of people consider to be flat speakers in a room. In this context, yes, general compliance with the curve WILL indicate a high probability of the headphones sounding alright.

And yes, the ultimate test of any headphone is personal testing, but you don't have to go to broken earphone extremes to prove that the Harman curve or measurements indicates good performance either. In an objective way, you can pretty much tell if a headphone has low or high distortion, resonances and peaks and, most importantly, you can tell if a headphone is bright or dark/muffled sounding just by looking at it's FR. Just look at the oval shaped hifimans, they all have exaggerated pinnas and high frequencies on GRAAS fixtures, and that correlates really well with the communities sentiment over them.

Measurements as a tool are even more valid and useful when correlated with your own personal experience. Rather than using a be-all-end-all approach to Oratory presets and the Harman target, you should see them as a canvas from which you can adjust to your own preferences while considering your own ear's quirks using EQ. Doing so with measurements in mind guarantees MUCH better results than just tuning by ear. So to say that "headphones measurements are worthless" is a very big exaggeration.
 
You didn't catch the point, there is no trend in nothing, measurements are worthless because your ear channel is unique, the '' ear '' from the measurement ring is not your ear, so.. it's worthless for you, is useful for someone who got the exact same ear as the '' ear simulator '' in the measurement rig.


headphones are kind of very very personal thing, sure you can tune to the harman curve the hadphone, but a '' defect '' in the harman curve can be a fixed by the ear channel
Only true if individual FR variation can't coincide with general preference. Put to test some headphones can have a generally appealing tuning despite significant individual variation. That's how the Harman was able to validate their model to predict listener preferences in 80% of cases. Id say subjective evaluation is prudent before saying measurement are useless, although measurements can be made worse than useless, toxically wrong if you overestimate their predictive power, based on arbitrary deviations of a smoothed target. In this instance I'm actually going a step further than your point about measurements being useless.
 
I believe that the reason many people feel that the low end of the Harman target is too elevated is largely due to the fact that, with many headphones, especially when the low end is boosted using EQ, the low end becomes significantly distorted, resulting in a buildup of harmonics in the mid-low range. The AUDEZE MM-500 has no distortion in the low end and its characteristics are flat, but even when the low end is boosted according to the Harman target, although it may indeed seem a bit too elevated, it does not result in the abnormal low end one might experience when adjusting headphones like the HD600 to match the Harman curve.
 
I would really like to see the soundstage and imaging in reviews, like rtings. Their frequency responses are clearly off for every headphone that I own, so having much better frequency responses than them plus also providing soundstage and imaging quality, would make the reviews here even more valuable and increase the popularity and name of Audio Science Review :)
 
Talking about soundstage in headphones is nonsensical to me. It’s like measuring biceps on bunnies.
Large cups can give off the effect of a larger sound field but yeah..looking at fx Rtings or similar sites/reviewers who are trying to conjure up some form of objective parametres regarding soundstage in headphones is not enlightening at all. It more than anything else feels like a merging of placebo and left-overs from the speaker world.
 
I would really like to see the soundstage and imaging in reviews, like rtings. Their frequency responses are clearly off for every headphone that I own, so having much better frequency responses than them plus also providing soundstage and imaging quality, would make the reviews here even more valuable and increase the popularity and name of Audio Science Review :)
I am also of the stance that "soundstage" heavily depends on the individual's susceptibility to certain audio queues. Some folks can hear a big difference in "wideness" between something like the HiFiMan Arya Stealth and HiFiMan HE1000se while for me, image width and all are expectedly identical since they are basically identical in shape and driver position. The only "width" I hear is the distance between the two drivers, so moving up from a Jabra Elite 85h to the Arya Stealth or Sennheiser HD 800 S for me only gains something like a centimeter in either direction. The number of times I've seen descriptions of a headphone having a "bigger" or "wider" soundstage when they could have just as well said that the pads were bigger and the drivers further, though yes, supposed differences between HiFiMan eggs is a weird phenomenon not apparent to everyone, whether or not dependent on volume-matching, else some being especially sensitive to differences in the level of the "HiFiMan dip". I seem to hear imaging independently of tonal queues for distance, though upper midrange dips like in the two mentioned headphones can at least add some timbral distance to vocals, but otherwise dulls some clarity.

For imaging quality, I would mainly look to the matching of the frequency responses between the two channels, but even that is subject to pad placement variations on your head or any asymmetry in your physiology. Some subjective imaging quality or "sharpness" may depend on the treble response of the headphones and/or the recording. To me, ideally, when panning a pink noise sound source from left to right, all frequency components should image coherently from the same point, whereby any drifts of some frequency bands left or right if not up or down imply physical or perceptual imbalances in frequency response or inaccurate interactions with your HRTF, whereby this may cause one to hear some frequency components of the same instrument or sound source imaging from different locations, which IMO is bad. E.g. Some stereo recordings that will image cleanly on a line between the two speakers or the two virtual ones in my binaural DSP may have treble content erroneously imaging above my head through almost any headphone due to the lack of proper HRTF reconstruction and lack of crossfeed; I guess some folks call this imaging error "holography" or "3D sound".

As for RTINGS' "soundstage" measure, it is fair to see how well a headphone matches the HRTF received by the speaker reference, but I'm pretty sure a perfect match by their metric would be like little more than EQing a match without crossfeed, and as far as I know, doing so will only somewhat match the speaker tonality, but cannot push far panned sound sorces at the left or right forward.
 
I am also of the stance that "soundstage" heavily depends on the individual's susceptibility to certain audio queues. Some folks can hear a big difference in "wideness" between something like the HiFiMan Arya Stealth and HiFiMan HE1000se while for me, image width and all are expectedly identical since they are basically identical in shape and driver position. The only "width" I hear is the distance between the two drivers, so moving up from a Jabra Elite 85h to the Arya Stealth or Sennheiser HD 800 S for me only gains something like a centimeter in either direction. The number of times I've seen descriptions of a headphone having a "bigger" or "wider" soundstage when they could have just as well said that the pads were bigger and the drivers further, though yes, supposed differences between HiFiMan eggs is a weird phenomenon not apparent to everyone, whether or not dependent on volume-matching, else some being especially sensitive to differences in the level of the "HiFiMan dip". I seem to hear imaging independently of tonal queues for distance, though upper midrange dips like in the two mentioned headphones can at least add some timbral distance to vocals, but otherwise dulls some clarity.

For imaging quality, I would mainly look to the matching of the frequency responses between the two channels, but even that is subject to pad placement variations on your head or any asymmetry in your physiology. Some subjective imaging quality or "sharpness" may depend on the treble response of the headphones and/or the recording. To me, ideally, when panning a pink noise sound source from left to right, all frequency components should image coherently from the same point, whereby any drifts of some frequency bands left or right if not up or down imply physical or perceptual imbalances in frequency response or inaccurate interactions with your HRTF, whereby this may cause one to hear some frequency components of the same instrument or sound source imaging from different locations, which IMO is bad. E.g. Some stereo recordings that will image cleanly on a line between the two speakers or the two virtual ones in my binaural DSP may have treble content erroneously imaging above my head through almost any headphone due to the lack of proper HRTF reconstruction and lack of crossfeed; I guess some folks call this imaging error "holography" or "3D sound".

As for RTINGS' "soundstage" measure, it is fair to see how well a headphone matches the HRTF received by the speaker reference, but I'm pretty sure a perfect match by their metric would be like little more than EQing a match without crossfeed, and as far as I know, doing so will only somewhat match the speaker tonality, but cannot push far panned sound sorces at the left or right forward.
At the end of the day you can only do so much when you're talking about two mini speakers strapped directly to the side of your head vs optimised speakers in a room - at least in terms of authentic soundstage/imaging. I mean if you go to the trouble & expense of Smyth Realiser and do all in the in-ear measurements and have access to an excellent speaker system set up in an incredible room for those in-ear measurements, then by accounts the Smyth Realiser will be able to reproduce that, although I've not experienced it myself some esteemed members have tried it and are wowed by how good it is. But yes, at the end of the day we're limited by what can be done by strapping two small speakers to the side of your head with regards to soundstage/imaging........but you can still get good tonality and excellent bass performance along with excellent resolution helped by comparatively low distortion......and still the imaging/soundstage in headphones can be enjoyable, but it's limited right.......it helps to have perfect channel balance throughout the frequency range though as an additional point to be thrown in.

EDIT: Although I do accept & experience that soundstage is better in some headphone than others. I've got an HD800 and that's about same as K702 in that regard (and I came back & edited this sentence because I listened just now to one of my soundstage tracks and K702 was better than HD800.....it's close between the two and will depend on your EQ & track) and I'm talking after both have been EQ'd to Harman Curve and also channel matched through the whole frequency range by measurement on my miniDSP EARS rig. And yes HD600 is my worst headphone for soundstage and also imaging - it's all close with left blob / middle blob / right blob, doesn't have smooth panning which is strange. HD600 does a lot right, but imaging & soundstage is not one of them in my experience. I can't really explain these differences in soundstage & imaging, but perfect channel matching does help, but is not the only factor. Still, these are fairly subtle differences in soundstage/imaging when I'm comparing say my EQ optimised K702 & EQ optimised HD800, with a somewhat wider gap down to HD600 and my other headphones just slotted inbetween, but it's gonna be pretty subtle for most people I think.
 
Last edited:
Until the methodology improves, I would rather trust my hearing as soundstage evaluation tool.

OTOH any time I heard something wider than average it came with a penalty of non-neutral tonal profile. So, I don't put much stake in soundstage, and I get suspicious when someone seems fixated on soundstage. It will make me wonder if such a person has reasonable priorities.
 
Last edited:
At the end of the day you can only do so much when you're talking about two mini speakers strapped directly to the side of your head vs optimised speakers in a room - at least in terms of authentic soundstage/imaging. I mean if you go to the trouble & expense of Smyth Realiser and do all in the in-ear measurements and have access to an excellent speaker system set up in an incredible room for those in-ear measurements, then by accounts the Smyth Realiser will be able to reproduce that, although I've not experienced it myself some esteemed members have tried it and are wowed by how good it is. But yes, at the end of the day we're limited by what can be done by strapping two small speakers to the side of your head with regards to soundstage/imaging........but you can still get good tonality and excellent bass performance along with excellent resolution helped by comparatively low distortion......and still the imaging/soundstage in headphones can be enjoyable, but it's limited right.......it helps to have perfect channel balance throughout the frequency range though as an additional point to be thrown in.

EDIT: Although I do accept & experience that soundstage is better in some headphone than others. I've got an HD800 and that's about same as K702 in that regard (and I came back & edited this sentence because I listened just now to one of my soundstage tracks and K702 was better than HD800.....it's close between the two and will depend on your EQ & track) and I'm talking after both have been EQ'd to Harman Curve and also channel matched through the whole frequency range by measurement on my miniDSP EARS rig. And yes HD600 is my worst headphone for soundstage and also imaging - it's all close with left blob / middle blob / right blob, doesn't have smooth panning which is strange. HD600 does a lot right, but imaging & soundstage is not one of them in my experience. I can't really explain these differences in soundstage & imaging, but perfect channel matching does help, but is not the only factor. Still, these are fairly subtle differences in soundstage/imaging when I'm comparing say my EQ optimised K702 & EQ optimised HD800, with a somewhat wider gap down to HD600 and my other headphones just slotted inbetween, but it's gonna be pretty subtle for most people I think.
The second part of https://www.head-fi.org/threads/anyone-into-crossfeed.961533/post-18068536 (post #78) is a free way to get a taste minus the head-tracking; for clarity, I prefer trying to simulate the anechoic sound with outdoor measurements and upmix in further than at-home practical first reflections later if I like, though I admit the greatest cost on my end was the overkill use of Genelec 8341A speakers; I've nonetheless had fantastic results this way. From what I remember, for PRIR capture, the Smyth-Realiser system uses a lot less samples than the https://www.earfish.eu/ method of HRTF capture, whereby I haven't had a chance to compare the efficacy of their interpolation method compared to the Earfish method that also captures granular elevation samples.

As an aside, I figured to take a closer listen at playing pink noise equally through both channels and feeding this into a volume panner in Reaper, comparing the HE1000se, Arya Stealth, Meze Elite, ATH-M50xBT, and Jabra Elite 85h, the latter two unbalanced and the rest balanced.
  • I generally had the feeling that the latter three gave me more sense of a vaguely frontal pan that I would almost call "good enough imaging" while the HiFiMans had things closer to my forehead if not slightly behind it, though later in the listening session, I started perceiving it more forward like with the ATH-M50xBT's treble. In all cases, the treble tended to image a bit high around forehead level while the midrange noise bands were more level.
  • With a pan law of 4 dB, which I've found maintains the volume best for panning with my speaker simulation DSP, the non-DSP pans while having the treble move left and right had the general feeling of lobing in volume level between the center and the front left and front right. The lower frequency noise bands also tended to lag behind and stay closer to the center with the highest frequencies being the most panned--consider that for traditional crossfeed like with https://bs2b.sourceforge.net/, I had found that it struggled more with pushing the higher-frequency components of far-panned sources forward than their lower-frequency components.
  • At the transition between 80% and 100% pan in either direction, I start to feel a kind of "quantum tunneling and superposition effect" with volume levels jumping between the upper front left or right of my head to being right in the middle of the drivers with 95% to 100% finally presenting the jarring void of having noise playing only in one ear.
  • With my personalized HRTF rendering DSP turned on, the pink noise coherently images from around one point cleanly along a line in front of me, moving left and right between the virtual speakers with a more clear sense of "looking at" the sound source, no frequency bands imaging high or low (in most cases) or leading or lagging, more coherent than anything I've heard with stock headphones. There is no jarring transition between 95% and 100% pan.
In short, when you hear proper coherent imaging free of room reflections, there's no turning back, and while I in this pink noise pan test have a chance of noticing some slight differences in presentation in new headphones I may encounter, for my ears, they would all most likely present imaging errors that I would want to correct such that comfort, distortion, and EQability would be my priority.
 
Last edited:
https://www.head-fi.org/threads/anyone-into-crossfeed.961533/post-18068536 (post #78) is a free way to get a taste minus the head-tracking; for clarity, I prefer trying to simulate the anechoic sound with outdoor measurements and upmix in further than at-home practical first reflections later if I like, though I admit the greatest cost on my end was the overkill use of Genelec 8341A speakers; I've nonetheless had fantastic results this way. From what I remember, for PRIR capture, the Smyth-Realiser system uses a lot less samples than the https://www.earfish.eu/ method of HRTF capture, whereby I haven't had a chance to compare the efficacy of their interpolation method compared to the Earfish method that also captures granular elevation samples.

As an aside, I figured to take a closer listen at playing pink noise equally through both channels and feeding this into a volume panner in Reaper, comparing the HE1000se, Arya Stealth, Meze Elite, ATH-M50xBT, and Jabra Elite 85h, the latter two unbalanced and the rest balanced.
  • I generally had the feeling that the latter three gave me more sense of a vaguely frontal pan that I would almost call "good enough imaging" while the HiFiMans had things closer to my forehead if not slightly behind it, though later in the listening session, I started perceiving it more forward like with the ATH-M50xBT's treble. In all cases, the treble tended to image a bit high around forehead level while the midrange noise bands were more level.
  • With a pan law of 4 dB, which I've found maintains the volume best for panning with my speaker simulation DSP, the non-DSP pans while having the treble move left and right had the general feeling of lobing in volume level between the center and the front left and front right. The lower frequency noise bands also tended to lag behind and stay closer to the center with the highest frequencies being the most panned--consider that for traditional crossfeed like with https://bs2b.sourceforge.net/, I had found that it struggled more with pushing the higher-frequency components of far-panned sources forward than their lower-frequency components.
  • At the transition between 80% and 100% pan in either direction, I start to feel a kind of "quantum tunneling and superposition effect" with volume levels jumping between the upper front left or right of my head to being right in the middle of the drivers with 95% to 100% finally presenting the jarring void of having noise playing only in one ear.
  • With my personalized HRTF rendering DSP turned on, the pink noise coherently images from around one point cleanly along a line in front of me, moving left and right between the virtual speakers with a more clear sense of "looking at" the sound source, no frequency bands imaging high or low (in most cases) or leading or lagging, more coherent than anything I've heard with stock headphones. There is no jarring transition between 95% and 100% pan.
In short, when you hear proper coherent imaging free of room reflections, there's no turning back, and while I in this pink noise pan test have a chance of noticing some slight differences in presentation in new headphones I may encounter, for my ears, they would all most likely present imaging errors that I would want to correct such that comfort, distortion, and EQability would be my priority.
I actually spent some time yesterday messing with the Crossfeed option in Neutron Player. I tried the 3 different presets, I think I preferred the Meier implementation which is option number 3 apparently. When flipping it on & off when listening to music sometimes I did sense that it moved a bit forward away from me in space when I flipped on the Crossfeed, but it didn't seem to reliably always do so. Crossfeed also changed the overall volume level slightly so it made it harder to compare. Strangely as I flipped back & forth between on & off Crossfeed it felt like I was about to develop a headache the moment I flipped to the other option (either on or off) and then that feeling would dissipate really quickly & I'd be back to normal. Overall though I liked it better without Crossfeed - it seemed to pollute the sound a little and make it less clean overall, but that might have been influenced by the volume changes too. I didn't particularly like it though anyway when I increased the volume to compensate.
 
I actually spent some time yesterday messing with the Crossfeed option in Neutron Player. I tried the 3 different presets, I think I preferred the Meier implementation which is option number 3 apparently. When flipping it on & off when listening to music sometimes I did sense that it moved a bit forward away from me in space when I flipped on the Crossfeed, but it didn't seem to reliably always do so. Crossfeed also changed the overall volume level slightly so it made it harder to compare. Strangely as I flipped back & forth between on & off Crossfeed it felt like I was about to develop a headache the moment I flipped to the other option (either on or off) and then that feeling would dissipate really quickly & I'd be back to normal. Overall though I liked it better without Crossfeed - it seemed to pollute the sound a little and make it less clean overall, but that might have been influenced by the volume changes too. I didn't particularly like it though anyway when I increased the volume to compensate.
I remember foobar2000's Crossfeed plugin ("component") being of the sort that felt like a mere slider between stereo and mono, thus compromising clarity, whereby it was only when I tried the linked bs2b (which albeit also incurs a volume level offset) that I finally got a then more convincing pushing of most of the image forward without much of the feeling of things getting congested, just that some sounds like panned trumpets or other higher-treble content would tend to remain close to my ears. Then proper HRTF DSP for me further maintains proper image coherence and separation compared to stock headphone sound, especially near the far pans, even not really stretching point sources when I move the virtual speakers further apart. If ever you have a chance to try out CroPaC + Reaper as linked.
 
Back
Top Bottom