I think I get what you are saying here. But aren’t you forgetting the additional gain caused by the interaction of your ear with the headphone? Like if you had a piano playing live, then you’d have your individual ear gain that your brain would normalize for, just like it has been all of your life. But the brain is using that same normalization when hearing the recorded piano playing through a headphone, and in that case there is additional gain from the headphone’s interaction with your ear which is dependent on the individual headphone. Isn’t this what is referred to as acoustic impedance? Or am I missing your point entirely?
I do think that there is a bit of normalization that the brain does with hearing over the course of a long listening session. Like when you use an equalizer to boost the bass a bit. After awhile, you don’t really notice it until you take the boost off again. I don’t see any reason why this wouldn’t also apply to small variations in frequency response. It’s basically the same thing.
I’m not quite sure what you mean. By additional gain do you mean from the headphone’s unique coupling interaction with your head? Because that may be the cause of some disagreement.
But otherwise, the normalization that everyone has will be different for each person (because the gain factors are also different). So it’s not as though ear gain (a) gets the same normalization as ear gain (b).
Think of normalization as literally the reason the world sounds ‘normal’ to each of us, in spite of the fact that we all have different ears with different gain factors. This on its own would require some variation in terms of how it gets normalized. But that also means that the same unique normalization occurs when you listen to a headphone. Remember that headphone measurements are taken as if the microphone were at the ear drum, so headphone measurements (when done on industry standard rigs) are effectively showing how the headphone’s frequency response interacts with the gain factors of the human ear. These gain factors may be slightly different from person to person, and that means the FR may look a bit different depending at whose ear drum we’re measuring - but because we’re all normalizing for these gain factors, in theory the difference between the measurements (of the same headphone on different ears) shouldn’t mean it sounds all that different to different people.
Again, I say ‘in theory’ here, because I’ve often wondered if there’s a limit to this. Like maybe one person’s ear canal is so small that it causes certain soundwave interactions that the brain can’t anticipate or normalize for. I kind of had my mind blown a little bit recently when Mad told me that sound waves can potentially take multiple paths to the eardrum, and this has made me think maybe this is responsible - probably not for the average ear, but maybe in edge cases.
Yes exactly. I don’t see how the brain’s normalization learned by listening to the world around you could compensate for this.
Another way to explain it might be that if we could somehow play the sound from the same position as the headphone, but in free space, I agree it would sound “correct” as any ear gain is compensated for by your brain’s normalization. But the headphone itself is interacting with your ear (and head) in ways that your brain is not used to normalizing.
Yeah I have some worries about this - like maybe there’s some interaction with the physical ear that causes a difference that’s not going to be normalized for - but then again because the pinna flange is only one small part of overall ear gain the effect would most likely be minimal. I think it’s also worth considering the effects of stuff like clamp pressure. There’s always a chance that this can significantly change the headphone’s FR for a given ear, and so how wide/narrow a person’s head is may also be responsible for why they hear something differently. So yeah I think variation for gain that’s the result of external factors like coupling likely plays a significant role in disagreement about how we hear a headphone.
At the end of the day I always come back to using the reviews I read as guides to what the Headphone ( or other Equipment ) is like. If I decide to commit to buying something I like to prepare myself for the experience to be Better Than / Same As / Not as Good as the reviews I read.
I’ve found with the Headphones.com reviewers they have fairly similar hearing to me so their reviews make good guides.
I know from reading experts (and you probably know this too) that the higher the frequency, the larger the effect of this and that above about 10khz it becomes significant enough that measurements in that region cannot be relied upon. I also know that newer measurement rigs claim to be accurate above 10 khz, which seems to be misleading to me.
Yeah this one worries me personally because I can’t stand wearing high clamp headphones. Although I don’t notice any change in the sound when gently pushing the earcups in on open back headphones. Maybe I’ve just been lucky in choosing headphones that are not sensitive to clamp or have self selected such designs because I don’t like high clamp. I would think that closed back headphones would be more sensitive to this than open back.
Some are more sensitive to that than others, for example I noticed quite a bit of variation with the HD560S depending on clamping pressure on the rig. But yeah with regards to higher frequencies not being as accurate, it’s definitely true that they have more positional variation, but that doesn’t necessarily mean they’re misleading. It just means you have to be more careful to get the positioning right on them - and for the 5128 it would be more accurate if you did that. I just find that it matters less above 10khz specifically because of the positional variance that exists.
But, I also just wanted to note that this isn’t necessarily the effect of interaction with the pinna flange. It could be just varied gain from all factors. Information above 10khz isn’t pictured here, but this is just to show how significant other factors are apart from the flange itself.
Yeah the position of the headphone is a factor but I was referring to the interaction of the pinna with the headphone when I said that variation increases with frequency, especially above 10khz. So the fact that any measurement device uses a specific pinna means that measurements above 10khz are not useful, even if the measurement itself is accurate (for that pinna). Because it is unlikely that those measurements above 10khz would apply to an individual.
The problem is that the standard deviation (“how much a specific individual deviates from the average”) gets higher at high frequencies.
So much so that at frequencies above ~10 kHz it’s virtually useless to specify an average.
What’s the average of 0.01 and 20.000?
Sure, it’s around 10k, but how meaningful is that to the individual? Not very.
Which is one of the reasons why I’m not convinced that the B&K 5128 will prove very usefull in the future.
Yes, it has specifications up to 20 kHz - but this only means that its behaviour is specified up to 20 kHz.
It does not mean that its measurement results will align with what you hear up to 20 kHz, because at 10k-20k, your ears will be very different from the 5128, just like they’re very different from my ears and everybody else’s ears.
Yeah this is something I completely agree with. Perfect accuracy above 10khz doesn’t really matter since you inevitably get significant positional variance regardless. So it might be ‘more accurate’ for how you would hear it at that position, but that doesn’t really help with getting a consistent, intelligible result. While I appreciate Jude’s efforts on the frontier of measurement accuracy (he has a great writeup on this), I’m inclined to agree with folks like Oratory and Crin that the 711 standard still has quite a bit of life left in it - in particular because we now have cross comparable databases going on.
My personal experience with BA IEMs suggests this may be the median case rather than edge case with at least in-ear products. Cheap BAs (e.g., KZ) are often very peaky, but also have abnormally localized treble. The localization experience doesn’t match any real-world environmental sound source other than mosquitoes in the ear, nor is it similar to any dynamic or planar driver. Regarding effects of ear size, the Etymotic sound channel is quite narrow and directional. Its design accounts for superior isolation, but the narrow channel may be responsible for the extreme sibilance I perceive [so I sold mine].
I continue to resist the assumption of a minimal impact, per my experience with ‘cupping’ one’s ears (or merely flexing the pinnae), as well as playing with stethoscopes and old-school trumpet hearing aids. A slightly different reflection here and there may add up to big downstream effects. A butterfly’s wings in China causes hurricanes in Florida…?
Agreed. To my ears treble variations are routine and either (a) become shrill/peaky with better quality transducers or (b) ‘brick wall’ into a hissy blob of upper-range noise that sounds a bit like a cymbal. PA systems/live concerts also often blob the treble.
This will obviously have an impact for how you hear things - your brain isn’t normalized to the additional effects that cupping your ears would induce. But again that doesn’t mean that in normal conditions, differing ear gain would be responsible for hearing the same headphone differently. Like, just to be clear I’m not saying there aren’t other reasons why we may hear things differently, just that in theory, because of individual normalization that occurs (in everyone), varying gain factors (from different ear shapes) aren’t responsible. And also, under typical circumstances, there should generally be quite a bit of agreement for how things sound - at least for over-ear headphones - specifically because of individual normalization that’s been operative for our whole lives.
This is also a very interesting subject. There’s a much wider variance in impressions for IEMs than there is for over-ear headphones (and you get some truly wild tunings with certain IEMs). I asked Crin about this and he had a very satisfying answer. With over-ear headphones, if the manufacturer is trying to get the headphone to loosely match what the brain expects to hear, they only have to assume (predict) the extra-aural gain effects (because we generally know what the rest would be for the average ear). And there may be additional factors for what sounds ‘normal’, like head, torso, angle of the listener and so forth - but also for what the expectation would be for the room (enter the Harman bass shelf just for fun haha).
For IEMs on the other hand, the manufacturer has to assume (predict) all of these factors, plus the multitude of variances that also exist due to the different pinna. So, while it might seem like bypassing certain gain factors makes things easier (in this case the pinna flange and concha), it also introduces an additional point of variation.
So for example, if the tuning assumes pinna effect (a), matching the ear of person A, but person B’s brain is normalized for pinna effect (b), then it likely won’t sound right to person B, even if it’s fine for person A. This kind of goes back to the previous discussion as well, where differences in impressions are more likely to occur as a result of physical interactions or unique couplings - and in the case of IEMs, it’s literally bypassing a physical part of the ear that our brains are otherwise normalizing for.
I’m not trying to be argumentative but while I agree positional variance has a large influence, I don’t think that is what oratory1990 is talking about in that post. He doesn’t mention position. He mentions that a “specific individual deviates from the average” and “your ears will be very different from the 5128”.
Yes! This is a BIG deal begin able to compare measurements. For one thing it might shed some light on unit variance of various headphones. And for another if (like me) you are interested in using EQ it gives multiple data sets to choose from.
Yes, I’m just meaning from conversations I’ve had with him. I should amend my post because you’re right this isn’t what he’s talking about in what you’ve quoted. But you can see the effects of positional variance on even the 711 rigs. By the way, there is a connection there. As you’ve noted, coupling differences (on different people), will influence the effects similar to that of positional or clamp pressure variance. The nice thing about testing by ear however is that you can generally get a sense of where it sounds the most appropriate - as in which position sounds best. That’s not so easy to do with just the rig, you have to kind of match it to what you hear when it’s at its best position.
Would it be useful to publish a sort of “best case / worst case” for high frequencies?
The idea would be to show the best/worst results from different positions on the test rig which would give a range of what different people could experience.
That may well require a norming study with a variety of human participants. A test rig is a model of ears based on an ‘average’ or ‘median’ human head/ear size, shape, and placement. It’s just a model.
A norming study would test specific frequencies with different people and headsets in different positions. The process would be a lot like going to an audiologist for a hearing test. And tedious. And costly.
After that one would know for sure what people COULD experience – then fold that back to the test rig (model) to validate its ability to cover the full spectrum of the possible.
I think when it comes to publishing measurements, we should be aiming for ‘best case’, because you can get some very strange results that would be immediately recognizable as wrong when listening.
I’m struggling now to make sense of drastically different conclusions in reviews. Is it differences in perception, differences in equipment/setup or just old fashioned people trolling?
Too many mutually-exclusive conclusions about the same headphones.
Edit: I don’t mean the official headphones.com reviews. I was talking about reviews from various different sites.
Yeah, the position and clamp force will vary with the individual just like ear shape. And all three together will influence (if not determine) the interaction between the headphone and your ears. I hadn’t thought of that.
I guess he did play a mean guitar at least…