[Rhyehr]

Questions:
I have searched online and can't find anything.
- What methodology (songs, headphones, EQ's) were used to determine the Harman Target? Why
- What studies determined Diffuse Field was optimal and how? All I can find is that it was concluded that you need ear gain for a headphone to sound like a speaker system in a good room.. but I don't see why that's our target for good audio?

Preface:

I've been on an audio kick for a while now ( A few years, ~ 30 iems/earbuds including TOTL units, Sennheiser over ears) and have found that there's only one iem that seems to have a very coherent soundstage - The Campfire Audio Honeydew. I've also noticed that the coherency of the honeydew has been mentioned by Head Fi reviews.

The only difference I could find that makes it stand out is that it has no "ear gain" (the boost between 2 to 4 khz), and a steady rise from mids to bass.

In the track "In the Distance" by Yosi Horikawa, I discovered when I remove the ear gain on other iems it also increases my ability to perceive footsteps as being on the ground, rather than right in my ear. Others who have tried it have found similar results.

This has lead to my current theory being that ear gain increases the perception/"resolution" of the 4khz region but it's so much as to throw off where items are perceived to be.

According to this logic, I could see gaming companies like HyperX having 4khz dips in their headphones on purpose since it actually helps the consistency of perception of auditory queues and hope audiophiles could come to accept that alternative "worse" tunings may be important for gaming.

 

[castleofargh]

Most work by Harman on their targets was done strictly for headphones(a dozen papers over about a decade). That's where you'll find all the questions, assumptions and how they tested them. You can get a lot of information even if you don't find the papers(AES paywall), by looking up Sean Olive on youtube and checking just about any video with him that's not about speakers.
Here is an article by Olive to try and summarize the headphone research they did https://acousticstoday.org/he-perce...uality-what-do-listeners-prefer-sean-e-olive/
I link it mostly because he does talk about DF and explains probably even more than you cared to know.

The IEM curve came only once the headphone research was done, and is a much simpler approach with maybe 20 trained listeners to establish the curve, and then some more people(maybe a hundred in total) to verify the resulting curve against other popular IEMs' FR. It's not nearly as involved, but it doesn't matter too much because even more causes for variations come with IEMs(insertion depth, seal quality having an even bigger impact than on headphones, and of course the choice of tips).

You don't need gain around 3kHz because it's something you get from your ear canal whether you like it or not. Any sound you ever got from the environment around you reached your eardrum after being boosted in that region by your own ear canal.
Headphones typically don't have much gain at that frequency. But then you ask, why do we see a big gain on the graphs? Well, the measurement coupler has an "ear canal" with that resonance to mimic how an average human ear would get it. The measurement rig is after all an ear simulator. My point is that if an IEM ends up flat with a RAW FR from a measurement rig, then it clearly means that there is in fact a dip in that frequency range and the coupler used for the measurement happens to compensate for it and give a flatter response. There is nothing in psycho acoustic suggesting that we should have no extra energy in the 3kHz range. But of course there is always a legitimate question about how much gain we really need.

From all that, 2 things are worth noting:
1/ The coupler will have a resonance somewhere near 2.7kHz(at least the old ones did) because that seems to be a good statistical average for human ears. But your own ear canal might boost closer to 2.5kHz or as high as 3.5kHz depending on your ear canal length. And that's just the frequency. The amplitude of the gain does also vary by several dB from people to people.
So you have some gain, and need it to feel like you're hearing normally(your idea of neutral response). But how much and precisely where that gain is for you, well that's something statistics and target curves based on averaging can't know. You're you.

2/When using an IEM, depending on insertion depth, the length/volume of your ear canal can be significantly reduced, meaning the gain from your ear canal's resonance will now boost a different frequency(a higher frequency). That mess ideally should be compensated by the IEM's tuning, but how can they know how long is your own ear canal, what tip you'll use and how deep you'll insert the IEM? Maybe the guy making the IEM will rely on his hearing so it sounds right to him, maybe he'll rely on statistics like the Harman curve in the hope of satisfying a bigger market. Anything is wrong for someone so nothing will be the preferred frequency response of everybody.

That leads us to your own experience. It's your own and different people can be, and often are, different when it comes to hearing. I discussed the 3kHz gain from the ear canal, but of course the outer ear and the head also alter the frequency response of every sound sources in our environment in a way that is specific to your own body(what you always imagined to be normal real sounds). Ideally, an IEM that tries to sound "natural", would wish to have predicted all those changes, accounted for the ones added by inserting the IEM in your ear, consider what was bypassed and what was altered, to come up with the right FR. To do that exactly right, you'd need to be used as a guinea pig while they figure out what you need. It's just not practical. So we're back to statistics and trying to make something that at least a group of people will enjoy.

I don't know about the IEM target, but the headphone target from Harman hopes to satisfy about 6 out of 10 listeners(by their own stats). If you don't like it, that does not prove something is wrong with you, that target, or their methods. It only suggests that you don't stand right in the middle with a most average human head and ears. You're with me and the remaining 4 out of 10. Not that rare or special.
From what you describe, it seems very possible that you need less gain in that region. But is it what most people need? After seeing a bunch of HRTF measurements, I can say with some amount of confidence that a recessed 4kHz isn't typical. Direction related changes in FR have usually more happening at 6kHz and above, with sometimes wild variations between directions and between listeners, but the 3-4kHz area is consistently reaching among loudest amplitudes of the audible range.
I do not know enough to draw conclusions on this. I'm sure it's not what will feel most natural to most people, but maybe when you attenuate 4kHz, you reduce the masking at 6kHz which could help with directional FR cues? I wouldn't bet my own money on that idea, but it seemed possible enough that I got it. Insert "it's something" meme. ^_^

 

[ADUHF]

Questions:
I have searched online and can't find anything.
- What methodology (songs, headphones, EQ's) were used to determine the Harman Target? Why
- What studies determined Diffuse Field was optimal and how? All I can find is that it was concluded that you need ear gain for a headphone to sound like a speaker system in a good room.. but I don't see why that's our target for good audio?

I think you may have a couple concepts slightly confused here, Rhyehr. Which is not especially surprising, given the complexity of these subjects.

For starters, ear gain isn't specifically related to approximating the response of a loudspeaker in a room. Ear gain simply exists because as sound waves pass from the outside to the inside of the ear, some frequencies are selectively enhanced more than others by the shapes and resonant characteristics of the pinna (esp. the concha) and ear canal. This is why there's usually a fairly large peak somewhere in the upper mids in the in-ear measurements of different kinds of sound sources, including diffuse and free field sources, and also the in-ear measurements of loudspeakers in a room, and most good over-ear headphones.

Diffuse field was the dominant model used for in-ear measurements before the Harman target came along, because most people prefer to listen to music in rooms with some reflectivity. And a free field (which is free of all echoes or reflections from walls or ceilings) doesn't really capture the ambient reflective characteristics of that type of space. The diffuse field model was never really followed with great precision though, because the sound in a typical room is more complex than either a spectrally flat diffuse or free field, and tends to favor more of the lower frequencies. The main thing that the diffuse field model lacked though was the room gain in the bass. And this is what the Harman target added.

The diffuse and free field curves, and Harman curve are all examples of transfer functions which describe the ear's response to different kinds of external sound stimuli.

 

[Rhyehr]

I think you may have a couple concepts slightly confused here, Rhyehr. Which is not especially surprising, given the complexity of these subjects.

For starters, ear gain isn't specifically related to approximating the response of a loudspeaker in a room. Ear gain simply exists because as sound waves pass from the outside to the inside of the ear, some frequencies are selectively enhanced more than others by the shapes and resonant characteristics of the pinna (esp. the concha) and ear canal. This is why there's usually a fairly large peak somewhere in the upper mids in the in-ear measurements of different kinds of sound sources, including diffuse and free field sources, and also the in-ear measurements of loudspeakers in a room, and most good over-ear headphones.

Diffuse field was the dominant model used for in-ear measurements before the Harman target came along, because most people prefer to listen to music in rooms with some reflectivity. And a free field (which is free of all echoes or reflections from walls or ceilings) doesn't really capture the ambient reflective characteristics of that type of space. The diffuse field model was never really followed with great precision though, because the sound in a typical room is more complex than either a spectrally flat diffuse or free field, and tends to favor more of the lower frequencies. The main thing that the diffuse field model lacked though was the room gain in the bass. And this is what the Harman target added.

The diffuse and free field curves, and Harman curve are all examples of transfer functions which describe the ear's response to different kinds of external sound stimuli.

To clarify, I literally created this post to ask for clarification since Audio Engineers seem mediocre at linguistics and documentation. I am definitely confused as to why speakers in a room was our target for a headphones FR for a long time, or why people trust the Harman Target when there's nothing most people can find on it's methodology (what music was used, how it was tested exactly, headphones used, etc.)

For example, Apparently the word intensity in audio engineering has a different definition than it does to the rest of the world. The word itself refers to "In extreme force" which in audio literally means decibels. But according to Oratory1990, intensity means something else entirely. Very confusing I must say.

Also, my main point is that I don't believe (yet) that it is as simple as boosting 4khz by 15 decibels, since your ear doesn't boost any frequency, it simply attenuates the sound.

Realistically what I'd like to know is the exact methodology used to A) Determine the Harman Target, B) What methodology was used to determine that "ear gain" was necessarily a thing and to derive the exact nature of how ear gain works.

By simple logic it would make sense to me that Bass passes through solid matter the easiest, therefore more bass reaches your inner ear, meaning a bass boost can be more accurate to real life. Past that I'd like to see information on how frequencies interact with matter but I can't seem to find many resources.

 

[Roseval]

I'm afraid you have to fork out some money as most of it is published as a AES paper.

http://seanolive.blogspot.com/2022/03/the-perception-and-measurement-of.html

 

[Rhyehr]

I'm afraid you have to fork out some money as most of it is published as a AES paper.

http://seanolive.blogspot.com/2022/03/the-perception-and-measurement-of.html

Thanks for the link, guess my search for Harman info is over because... no, I'm not paying for that. I don't care about what sounds good to most people I suppose, I care about replaying what a microphone records as accurately as possible to what my ears would receive irl. As a gamer my only question is "Can I perceive where this is accurately" and as an, I guess, now audiophile, I'd believe that accuracy in the perception of audio queues is the most pertinent part to an auditory experience.

I think further studies should be about how accurately sounds are perceived. Myself and others have found that in tracks like Yosi Horikawa's "In The Distance" and with audio in games, the footsteps seem much more accurately positioned around you when you remove alot/all of the ear gain. Which also makes my theory as to why HyperX and pro gamers tend to trend towards a bass boost/treble reduction, and why audiophiles are misguided in the idea that gaming headphones are "bad".