In Quest Of Neutral Headphones [Reviews / Measurements]

Discussion in 'Audio Science' started by Magnetostatic_Tubephile, Dec 30, 2018.

  1. b0dhi

    b0dhi New

    Joined:
    Nov 20, 2019
    Likes Received:
    32
    Trophy Points:
    13
    Location:
    Sydney
    Yes, but because headphones bypass most of the HRTF related effects that the brain compensates for, the same headphone will sound very different to different people.

    There is a way of doing what OP is trying to do though. Instead of using the MiniDSP EARS, you can use an ear canal mic and measure your own HRTF with a neutral set of speakers. That would be your target response. Then, instead of measuring the headphones on the EARS, measure them on your own head with the canal mic. At that point you can choose from a couple different options: either use DSP to make the headphone match your neutral target response, or you may want to minimise DSP artifacts, in which case you can keep hunting for a headphone that comes reasonably close to matching your target response and if you get lucky and find one, a small amount of EQing may be enough to get you to a neutral response.
     
  2. I_want_all_the_tacos

    I_want_all_the_tacos Friend

    Pyrate
    Joined:
    Oct 3, 2017
    Likes Received:
    440
    Trophy Points:
    63
    Location:
    San Diego, CA
    @b0dhi is correct about this.

    I think there is some confusion surrounding what "neutral" in headphones means and it can take a while to wrap you mind around. I'm a computational neuroscientist and a lot of my job is understanding how the brain inputs external stimuli and converts that to electrical impulses that create the reality we perceive. I've written about this idea of neutral and EQ in the past in other forums. But I think it might help to explain it with a visual analogy. This is sort of complex, and I just came up with it, but stay with me.

    Imagine we gather a group of people and show them all the same photograph. We can guarantee each person has the same visual input (the photo) and we know exactly what colors and shapes are physically there. But we don't know exactly what happens after that and how each brain is perceiving and forming the image in each person's mind. Basically that age old philosophical question of "do you see the color blue the same way I see blue?" But that's not important because in this case we know everyone is seeing the same photo and has the same external input. We don't care how their brain processes it. This is just like speakers that measure flat in a room. We know the exact same soundwaves and frequencies are reaching each person exactly the same. And how your brain individually perceives that is your own thing. But let's consider what happens with headphones.

    Imagine while showing each person that same photo we were able to put a tiny electrode inside each person's eye and record the exact electrical impulses each person's retina was sending through the optic nerve (that is how our brain "sees" and forms that image in our mind). We know because of individual anatomy differences that this electrical impulse response pattern will be different for everyone, even though they are all seeing the same photo. Next, since we hypothetically measured the retinal electrical impulse of each person, we can find the average electrical impulse of the group. If we take the real photograph away, and instead put a tiny electric stimulator on each person's optic nerve we can send that same electrical impulse to each person. If your optic nerve happens to work exactly the same as the average of the group's, your brain will "see" and form the exact same photograph in your mind (you could literally close your eyes and still see the photo perfectly in your mind). But if your optic nerve happens to not be exactly average, you will actually form an image of a photograph in your mind that is slightly different than the real one. It will be distorted. This is what happens with headphones and FR measurements. It is like bypassing the real, external photograph and instead injecting the downstream impulses into your eye directly and sending it to your brain. If your ear happens to be anatomically the exact same as the GRAS/HATS/MiniDSP ear that the headphone was measured on, you can indeed EQ it in a way that will sound the same as a flat measuring speaker. However, if your ear is shaped differently, then you will hear (and your brain will perceive) something completely different. Only if you know your own individual HRTF can you compensate for it to match a specific target FR (flat speaker, harman curve, etc). People that think they can just take an innerfidelity, Oratory1990, Rtings, etc. FR measurement and apply EQ to make it fit a specific curve and then actually hear the equivalent of what a speaker in a room is doing are wrong (unless of course you have the exact same ear shape).

    From the post I linked to above, I left this valuable resource:
    "Next, I urge you to watch David Griesinger’s lecture on binaural hearing and equalization . The link is queued up to the most important part of his talk, though I recommend watching the entire thing. He has calibrated 10 different peoples’ HRTF response so he knows how each person will perceive sound from a given headphone, and he then played each of them pink noise using HD600. He then demonstrates with samples exactly how each person hears the same pink noise being played to them and you can hear how dramatically different each person’s response is. It is a very striking and concrete example of understanding why each person should be using EQ to precisely try and hear what the original sound engineer had intended."
     
    Last edited: Nov 22, 2019
  3. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I'd argue that headphones don't bypass "most" HRTF, or as much as we think it does, i.e., creates a new individual headphone HRTF as to be different that the inverse HRTF in our brains doesn't work from person to person.

    It's not black and white. From playing around with different measurement rigs with ears, this is what I would posit in regards to headphones:
    • Concha influence on HRTF is not bypassed or changed much. The inverse HRTF for the cocha in our brains will still be at play. For IEMs, it will be different since everything except ear canal is bypassed.
    • Pinna effect tends to vary a little bit more, and it depends upon headphone. It's been my observation that planar headphones tend to yield more consistent compensated measurement results from different ears than dynamic drivers. I suspect dynamic drivers have stranger interplay with individual pinna to throw off inverse HRTF in the brain.
    • Variation of HRTF from person to person can be off quite a bit, but it's not hugely off, unless you have huge pinna like Alan Greenspan. There's still a bell curve.
    • The frequencies most affected are going to be 2-5kHz.
    • Head and torso effects on HRTF are insignificant.
    Bottom line is that it's going to be impossible to use any one measurement rig and compensation to determine a perceptive neutral curve for all humans. However, the differences are not going to be wildly different since most human beings ear structures are similar. And if there are any differences, they are going to be in the 2-5KHz region, with planars sounding more consistent.

    I could write a paper on this, but I don't have Sean Olive's job, and I dislike academia for the sake of academia. Note that any data that I have is very limited at this time.
     
    Last edited: Nov 22, 2019
  4. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    No.

    Explained above. It may sound slightly (a few db) different and mostly in the 2-5kHz area. And even then the differences in perception will likely follow a bell curve. And then then, it's going to depend upon headphone with some headphones sounding more different to different people than others.

    Weirdos 1 STD off the bell curve, unless they can admit so, are not allowed to post impressions on SBAF. :)
     
    Last edited: Nov 22, 2019
  5. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    This does not dodge the qualia issue in the "colour blue" problem. Your photodiode matrix would register an upside down image, yet the brain rotates it so it coincides with other sensory inputs. There have been experiments with people having to wear image flipping glasses and having this function of the brain turned off so they'd have an upside down image without the glasses. Naturally they would adapt in time and get better, once the glasses were off for some time.

    The qualia problem is inherent in the black box concept of the mind/brain. As in - there is no certain way of you to know how it feels like hearing Slayer to me and furthermore - you can't really know whether I have consciousness like you have. Maybe I'm just a meat robot... You could kinda neuroscience your away around it by matching EEG's to sound stimuli, but for that you'd need to be reductionist, which is boring af.

    This very much matches my experience. At SW we had 5 audio engineers tune measured responses, so they'd match to what they're hearing in the reference. There were headphones where positioning induced too much shifting so we'd refrain from releasing a correction profile. But generally the team would reach a conclusive curve every time. The upper mids were certainly the point of most discussions, but again - the differences were 1-4dB.

    The "everyone hears different, because everyone's ears're different" school of thought grinds my gears every time. Mostly because it's a very simplistic mechanic approach to psychoacustics. Do asian people see differently because of the shape of their eyes? Do you constantly see your nose despite that it covers a good portion of your visual FOV? No. Because you have a super complex computer between "you" and your earholes and eyeballs. And none of our senses are made for perceiving the world "as it is", they've evolved to present the world "as it's useful".

    Here's some interesting reading - https://medicalxpress.com/news/2018-03-brain-elevation.html
     
  6. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    When you sit people down at a meet and make sure everything is controlled, then talk about what you hear, and assuming you know the people well enough to calibrate for their own preferences, you do get a sense of personal "hardware" being at play. I'm positive some of it will be how their ears change driver interaction, but age (or other hearing damage/loss), hair, whether they wear glasses, head shape features and how it affects seal, etc. play a role as well.

    Now, take hair, glasses, and hearing loss out of the equation, and controlling for everything else you can, I'd say there's about...80-90% agreement in how different people hear the same headphone. And it's not usually the broad stroke stuff where people disagree, i.e. someone hearing a headphone as terribly bright and another hearing it as super bassy and dark, but more subtle stuff. Like the sort of changes you might get from light damping mods on a headphone. Nothing earth shattering.

    Personal preference, combined with what music they listen to in private, are the things that drive where most people draw their line on what they ultimately like the most or not.
     
  7. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Like what I said about the recent level matched blind tests at the Schiitr:

    Put people in a room with the same gear, same amp, same transducer, same source, same recording, same power, etc. Even let them take notes and not share, and only reveal results 10 minutes later.

    60%-85% will prefer the same things. And after accounting for personal preferences, it will be discovered that 85%+ heard the same. Hearing the same isn't the same as same tolerance or same preference.

    The only time I can buy the argument that people hear differently is with IEMs, and this seems to be from 3-6kHz where the concha has the most influence. Makes sense since IEM bypass concha.

    Otherwise the argument that all people hear differently is bullshit. This is why people on SBAF are sometimes asked to name associated gear, recordings down to specific masters, DAC filter settings, etc. The "discovery" process for why people heard something different is often illuminating.
     
    Last edited: Nov 22, 2019
  8. I_want_all_the_tacos

    I_want_all_the_tacos Friend

    Pyrate
    Joined:
    Oct 3, 2017
    Likes Received:
    440
    Trophy Points:
    63
    Location:
    San Diego, CA
    I absolutely agree with this. That's all my post was about, just to give a weird analogy to illustrate why speaker measurements and headphone measurements are presenting different things. And I also agree that theoretically all anatomical differences should be normally distributed and the average ear shapes used in the measurements rigs should be close enough that while there are variations person to person, the net result will be small, probably insignificant for the majority of people. I didn't mean to sound like we are all hearing vastly different things. Though in this hobby small variations in perception can mean a lot to people.

    Agreed, but that's why I said it doesn't matter in my thought experiment. What we each do individually doesn't matter if we are both physically shown the same thing. How you see blue in your mind could be how I see red. But as long as we are both seeing the same blue image, we both know how "blue" looks to us and we can both agree we are seeing blue. Cool study btw, the neural plasticity of the brain and our ability to adapt our perceptions to physical changes is always interesting.

    Anyways, didn't mean to come off as contrarian here. My own experiences and feelings align with what you all are saying, in real world when you gather a group of people more often than not there is a strong agreement about gear. And particularly when comparing gear, blind testing, whatever, since that is a relative comparison, it doesn't matter what your own brain does or how messed up the shape of your ears are, the delta between things should be the same for all of us and it should be clear if one thing is brighter or bassier than something else.
     
  9. b0dhi

    b0dhi New

    Joined:
    Nov 20, 2019
    Likes Received:
    32
    Trophy Points:
    13
    Location:
    Sydney
    Here is another measurement with fewer people which shows a clearer view of how much each individual differs:

    [​IMG]

    So yeah, that's a 30dB difference person to person out of only a few people, and there's no meaningful "bell curve" above 5kHz. It's about as unique as a fingerprint. The fact that people tend to prefer certain equipment is probably due to tuning below 5kHz and darker presentation above that, causing offense to the fewest number of people. Above 5kHz, all bets are off and there's no meaningful "generic" response to tune against.
     
  10. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,961
    Trophy Points:
    113
    Location:
    Philippines, The
    @Magnetostatic_Tubephile Moving here to keep the RAD-0 thread less messy :p

    Apologies if miscommunicated, but I didn't mean to imply that the 10dB trough between 2-6kHz was in any way measurement compensation error so much as it was just a difference in... choice of voicing? This voice is something Audeze cans generally seem to target since loads of them approximate it based on measurements and my limited experiences; while it is lacking in some regards (because perception of overtones of many, many things are heavily reliant on this region IMLE), it's pleasant enough that these headphones have found a degree of success in a turbulent battlefield like the headphone industry, as you'd noted.

    I agree that calling it a "reference" sound is bullshit, haha. I believe that the only reason they put that questionable quote up on their site is to appeal to a wider audience that want "studio neutral" sound even though they might not necessarily know what that is. No more reference than the HD800 is, though as I mentioned before I don't personally perceive that trough as disturbing timbre as much as it does spatial cues. It's part of why I like my HP-3s— the fake headstage is great for when I'm just plugging into my phone and listening to stuff on Spotify in bed!

    As far as I'm concerned, I agree that the quest for ~n e u t r a l~ sound is one that has many branching paths just because of how messy systems can be. Look at how AudioQuest threw the baby out with the bathwater voicing the Nighthawk! For it's great technicalities, loads of that was buried under overly warm sound because Skylar Gray believed that headphone designs shouldn't account for individual HRTFs/HATS transfer function. I haven't the same insight into Audeze's voicing because they didn't make as big a deal of it as AQ did in their marketing spiel, but I assume this wide FR depression is in part due to a difference in approach much as anything else, if it isn't plain euphony and preference.
     
    Last edited: Dec 3, 2019
  11. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    What exactly are we looking at here? Is this one headphone measured with an in-ear mic on multiple people? It's interesting to me because I measured my HD800 and my HD600 on me and 3 or 4 other people and up to 5kHz the response was very similar like in your graphs and the main difference was that my ears had 10dB less 6kHz than all of the other ears for both headphones.
    Obviously that doesn't show the complete picture, I feel we'd also have to measure the HRTF for speakers for that, but maybe the idea is to design a headphone that'll excite resonances in a similar way to speakers for most people. I know speakers at a 30° angle tend to excite an 8kHz resonance with my ear shape and headphones without super soft pads tend to aswell. I also think this is why headphone treble will never sound as natural as it should - the earcups will always cause some unnatural resonances. Maybe a relatively soft driver material and a special, less plane, shape could work the best here.
     
  12. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,961
    Trophy Points:
    113
    Location:
    Philippines, The
    @b0dhi and just to add to Serious's question, do those squiggles also account for hearing loss due to age and damage? Those sharp notches in the lower and mid treble region seem pretty suspect.
     
  13. m17xr2b

    m17xr2b Friend

    Pyrate Banned
    Joined:
    Mar 26, 2016
    Likes Received:
    3,988
    Trophy Points:
    93
    Location:
    United kingdomland of fish and chips
    This is why I believe MySphere is one of the best cans on the market
     
  14. wormcycle

    wormcycle Friend

    Pyrate
    Joined:
    Aug 13, 2016
    Likes Received:
    1,506
    Trophy Points:
    113
    Location:
    Toronto, ON, Canada
    I am not sure I understand.,either your comment or what the graph represents. What is listening is a microphone not a person. The differences on the diagram can only be related to what happens in your ear before you actually hear anything.
    Any part of human anatomy that is responsible for age based hearing loss is behind the eardrum, a microphone must be in front of it. Right?
     
  15. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,961
    Trophy Points:
    113
    Location:
    Philippines, The
    No you're right, I was being dumb about the method through which the data was gathered. I figured there was an element of prompting whether people could hear certain tones, as per a standard visit to the audiologist— apologies for the brain fart :p

    Do note I have no clue where in the ear system the mics were placed— I assume from the shape of the curves that they're about partway through the ear canal, though not necessarily at eardrum-point (cuz how the fudge would you even manage that without butchering the membrane).

    That said, interesting how there are massive nulls in most of the respondents' measured HRTFs. I know from experience and anecdotal evidence that sharp "cuts" in FR for headphones and speakers may not necessarily be audible, but I'm genuinely curious how the physiology of a person's auditory system can have big ol trenches like that. I'm guessing it's nothing the brain can't compensate for of course, but this could explain why some of my friends find the TH-900 to be less sharp than the HP-3 or TH-X00EB in the treble.
     
  16. Magnetostatic_Tubephile

    Magnetostatic_Tubephile Friend

    Pyrate
    Joined:
    Dec 27, 2017
    Likes Received:
    957
    Trophy Points:
    93
    Location:
    GoogleChrome, EU
    NOTE: While researching headphone market to find some new headphones for reviewing, I stumbled across a subcategory of wireless headphones incorporating DSP techniques to digitally linearize or otherwise enhance their sound. Finding the general idea intriguing, I carefully picked two headphones that seemed to approximate theoretical neutrality the best, based on available measurements. Do these products perform good enough to stand a chance against high quality passive headphones?

    Bose QuietComfort 35 II (stock)


    [​IMG]

    [​IMG]

    MEASUREMENT COMMENTARY:

    • Bose QC 35 II sounds fine in wireless configuration but using cable yields better fidelity. Also, only active mode provides neutral-enough sound to warrant inclusion of the headphone in this thread. Therefore, wireless and even moreso the passive mode were omitted from the review.
    • All active noise-cancelling modes (Off, Low, High) are measurably and perfeptually very similar to each other which was a very positive surprise. 'Off' mode appears a little bit more pure timbrally while 'Low' / 'High' ANC modes give a bit more depth, roundness and focus to headstaging.
    • Even though the slight measured channel imbalance can be spotted, I did not find it bothersome at all. If in need to tackle it, you can either slightly adjust the headphones on your head, or activate 'Low' / 'High' ANC mode. (DSP incorporated with the ANC modes seems to apply some volume levelling trickery.)
    • While bass frequencies are emphasized, perceived bass softness makes it less of an issue. In practice, speaking of bass presence, QC 35 II sounds closer to Auteur w/ Eikon pads than to truly bass-heavy headphones.
    • Mids and treble both feel flat and present to me, albeit a bit rough at times. Careful placement on your head is needed to avoid 2.5 kHz peak.
    • Even if you avoid the peak, distortion graph seem to correlate with subjectively heard tendency for shoutiness. That might actually be a plus to some listeners since it gives the Bose a vocal-centric presentation.

    [​IMG]

    HIGHLIGHTS:
    1. Probably the most tonally correct "active" (or at least ANC) headphone currently available on the market.
    2. ANC modes rivalling non-ANC mode in overall sound quality and engagement.
    3. Great ergonomics - lightweight and comfortable, high quality materials used all around, very easy to pair and operate in wireless mode, world-class active noise-cancelling performance.

    WHAT TO COUNT WITH:
    1. Cannot rival high quality audiophile headphones in timbre and technical proweness. Softness, anemicness, graininess, greyness, lack of dynamics, syntheticness and other terms typically used to describe sonically-compromised headphones apply here.
    2. Can be a bit too bassy and shouty tonally.
    3. Hiss with higher digital volume.
    4. Non-perfect ergonomics - level of ANC resets to 'High' each time you turn the headphone on, level of ANC cannot be set while in wired mode (you need to disconnect the cable first), stock cable is too short for desktop listening.


    Jabra Elite 85h (stock)

    [​IMG]

    [​IMG]

    MEASUREMENT COMMENTARY:

    • Jabra Elite 85h sounds fine in wireless configuration but using cable yields better fidelity. Also, the headphone automatically turns on when put on your head, making it impossible to operate in passive mode. Therefore, wireless and even moreso the passive mode were omitted from the review.
    • Active noise-cancelling mode is certainly listenable but considerably degrades sound quality. On the other hand, it helps to even out midbass to lowmid area, shifting tonal balance of the headphone a bit more towards neutrality.
    • Channel balance is perfectly fine when appropriate seal (see the graphs above) is achieved. This might be problematic though, as the headphone gets too bassy with perfect seal and too wonky with not-good-enough seal.
    • Unlike QC 35 II reviewed above, the Jabra are clearly bass-oriented headphones. Their bass is pronounced, punchy and 'warm'.
    • Despite the bass, Elites do not go too far from neutrality thanks to their linear mids, followed by reasonably balanced upper mids and treble. Some of the measured treble emphasis probably accounts for somewhat pronounced zinginess ("tsss" vocal sibilance, emphasized cymbals, ...).

    [​IMG]

    HIGHLIGHTS:
    1. Great ergonomics - lightweight and comfortable, high quality materials used all around, very easy to pair and operate in wireless mode, diverse active functions offered to users, water and sand resistant.
    2. Interesting blend of overall balanced sound signature with strong bass.

    WHAT TO COUNT WITH:
    1. Cannot rival high quality audiophile headphones in timbre and technical proweness. Softness, anemicness, graininess, greyness, lack of dynamics, syntheticness and other terms typically used to describe sonically-compromised headphones apply here.
    2. Too bassy to be even considered close-to-neutral.
    3. Hiss with higher digital volume.
    4. Stock cable is too short for desktop listening.

    TOP CONTENDER (as of 2019-12-27): NO...
     
    Last edited: Dec 28, 2019
  17. Magnetostatic_Tubephile

    Magnetostatic_Tubephile Friend

    Pyrate
    Joined:
    Dec 27, 2017
    Likes Received:
    957
    Trophy Points:
    93
    Location:
    GoogleChrome, EU
    SIDENOTE: I still feel that Harman-elevated bass is too much to be called realistic. Given a high quality driver (planar magnetics, Auteur or similar), you dont really need to boost the bass by 5dB to feel it. The experience with Bose makes me wonder which crappy headphones they used at Harman when developing their "neutrality" curve.
     
  18. dematted

    dematted Friend

    Pyrate Banned
    Joined:
    Dec 7, 2019
    Likes Received:
    2,161
    Trophy Points:
    93
    Location:
    Brooklyn, New York
    I agree with this. I've noticed that headphones tuned (or EQ'd) to the Harman Curve sound a bit bassy for me, to the point where I feel that they bleed into the mids. Of course, this could just be a result of the quality of the headphones I'm using, though I have come across a pretty substantial bloc of people who agree with me that the Harman curve makes bass a bit too prominent. I'm tempted to say that people who tend to be more into audio are, on average, somewhat more bass-shy in their listening tastes than the "average" subjects who the Harman curve was designed for.
     
  19. crenca

    crenca Friend

    Pyrate
    Joined:
    May 26, 2017
    Likes Received:
    3,825
    Trophy Points:
    113
    Location:
    Southern New Mexico
    Thanks for the measurements. The Bose QC35's are my workout & plane riding HP's. I hear them as a bit V shaped (but not unpleasantly so), maybe that 4-5 kHZ hump gives me this impression. They are also somewhat "synthetic" and ,even more, "grainy", but again for the price point and purpose not unpleasantly so. I would say they are even a "fun" sound...
     
  20. Stuff Jones

    Stuff Jones Friend

    Pyrate
    Joined:
    Feb 3, 2016
    Likes Received:
    1,790
    Trophy Points:
    113
    Great thread.

    I notice no Focals. Is this an intentional omission?
     

Share This Page