Touched by an IEM guru, running the SBAF Dick Gauntlet, and why target curves for IEMs won't work

@Marvey the HRTF from the outer ear is different in the nearfield with headphones compared to speakers. I posted measurements somewhere on here a long time ago that show that this is a major effect. Open headphones essentially "bypass" a large part of the outer ear simply because they're in the nearfield of the ear shooting the sound straight into the ear canal. With headphones we all have a very similar and only slight amount of gain from the outer ear. With speakers the curves vary far more. I have measurements somewhere on here for that too.

When it comes down to Hands' measurements, it's mainly because the ear canal is bypassed, which itself provides a shit ton of gain in the upper midrange. Do the same thing with a dummy head and the results wouldn't be too different.


EDIT: I'll just leave this pic from innerfidelity here for now:

 

"The head is designed to act exactly like the average human listening system. Inside the head at very much the same position of your eardrums at the end of your ear canal, are two very special microphones (one in each ear) that mimic the exact acoustic impedance characteristics of your eardrum. The shape of the ears is also very precisely defined (IEC specification 60268-7) so as to act just like the average human ear." - Link

(I assume this is what you're referring to: http://www.superbestaudiofriends.or...-hear-different-thread.2535/page-2#post-67954)
I'm fairly certain those were measured with speakers at quite some distance to the ear. The effect here will be similar to what I showed with my Riva S here in that more distance increases the outer ear gain effects.
Those were not the measurements I was talking about, but the others looked similar.

My philosophy is a bit different regarding voicing for speakers/headphones/IEMs:

For speakers I think flat in an anechoic room with smooth power response is good. I used to think that a upper mid dip like what Raidho does is good (but less strong), but to me it's more a bad way to try to cover up a bad room (With a mono speaker it might be a different story.). My speakers measure pretty flat at the listening position and it seems that a lot of the in-room measurements from stereophile are similar, not with a B&K curve. But that's really a discussion for another thread.

For headphones I feel the brain knows that something is up. There are peaks related to the ear that the brain doesn't know. The brain freaks out and doesn't use its inverse HRTF for the outer ear, but still corrects for the ear canal. The soundstage gets messed up and things sound like they're coming from inside of your head. The ear doesn't hear any reverb from the room and moving your head doesn't change the tonal balance. But there are still reflections between the driver and the ear. The brain can tell that the sound source is very close to the ear and it sounds like it.
For IEMs it's similar, but now the ear canal resonances are different, too. I feel IEMs need even less of a HRTF compensation (as measured at the eardrum level).

I don't mean to suggest that we should aim for a flat at the eardrum FR with headphones and IEMs, but flat at the ear canal opening generally matches my subjective impressions for headphones very well. And for IEMs I like to see only about a 5db bump on the IEC couplers centered around 2-3kHz, instead of 15db centered around 3-4kHz.

 

That particular paper @james444 brought up mentions the differences of model heads to real ears. The writers had wrote a few AES papers about the differences, measured results and binaural listening. Subjects found median plane localization, the part that depends most on correct directional FR, significantly more difficult with dummy heads than sticking a microphone in a random person's real ear.


Taken from this paper. Initials refer to the human subjects being used as "dummy heads." Median plane error would suggest pinna shaping at fault.

His version of the HMS head has the 711 couplers installed, which intends to simulate the impedance of the eardrum and attached middle ear.
As a side note, the ear canal has no serious contribution to directivity of the ear.

Seems reminiscent of the line of thought that lead Günther Theile and others to diffuse-field equalization 30+ years ago, but he assumed only the directivity of the ear was negligible. If you assume the total contribution of the pinna is minimal, that leaves you only with the transmission line-type behavior of the ear canal, first resonance below 3 kHz, next resonance around 8-9kHz, and so on. Headphones mounted on real ears and model ears show more gain between those resonance points.

Last I recalled, @Hands didn't stick his microphones right next to his eardrum, so the drum response wouldn't be measured.

 

What if they just want to get Schwifty in here?

Ill enlighten you all too after I get my doctorate. Wait 3 years and Ill have my bachelors, and Ill be able to enlighten everyone a little bit, but not fully...

 

@Marvey why exactly was this thread seperated?

With regards to the topic, a lot has been said here already, nothing new. I think at the moment, some things are not yet fully understood.

Eardrum measurement = eardrum measurement.
It isn't. If that were true, you could easily make one headphone sound exactly like another, or an IEM sound like a headband headphone. I've worked with people that were convinced the problem was so easy and linear, but it isn't.

The best example for me of the falsehood of the huge 2-3 kHz bump is the IE800. It has no bump to speak of really, so you would expect it to sound horribly laid back. It doesn't.

Its interesting to me at least that the 2 largest more recent studies for headphone target response using people both arrived at lower mid range bump than the HRTF style bump (Harman & Philips). I don't think it's 15 dB but I also don't think it's 5. From my own experience reverse EQ-ing an HD800, I landed blindly on 12, for whatever that's worth.

Pinna contributions
I think we need to split this into two things:

• Timbre, or simply put FR
• Spatial cues

No headphone without DSP does spatial cues correctly, end of story. There's no room, no reverberant field, no reflections, no crossed, nothing. Stop.

With respect to headphone FR measured on a dummy or human, there's not so much Pinna interaction o believe. The concha bowl, perhaps yes. But reflections of the outer edges like in a real environment? I don't think so. Those drivers get pretty directional for higher frequencies and are pointed directly at the ear canal (when angled).

Still, some things are not fully understood, as far as I know. (1) why do open headphones sound more "open", (2) why do IEMs appear to need more bass than headphones.... I've read peoples opinions about these, and I have my own, but I've yet to see hard proof.

Rant over.

Edit: I just had a thought that could be an interesting experiment for the Kemar/hats makers out there. Construct ears with more / less features to see which influence HP measurements.

 

Hey!

On one of those dummies, ok, hair.

Jokes aside, thanks everybody for the intersting discussion.

 

really? can you point me in the direction of at least one of these papers where people have molded ears with more or less features and measured the influence on the headphone response?

 

yeah, that middle link is a starting point. But pinna vs. no pinna is too coarse. I would be interested in seeing how different parts of the outer ear contribute to the measurement, if at all. My gut feeling (sorry) would think that a smooth pinna shape with a concha bowl would get you very close to an actual ear shape in headband measurements.

 

Probably should have no head hair, ample back, ear, nostril hair and man-boobs, to accurately represent the HRTF of typical ORFAS members.

You can custom order with any face you want, but I guess you might need Jude's consent for that

 

@Serious can replace Sean Olive, and I can replace the dummy.