A New Curve...

[TronII]

After learning about speaker orders, in combination with being exposed to some recent threads, I came up with a new idea for developing a compensation curve for circumaural headphones: I call it the "Open Air Response" (OA).
In a flat open field, with little to no background noise, (like the California salt flats) you place two first order speakers (so that the sound travels bidirectionally) with a known frequency response, and compensate them with an inverse curve, making them linear. After that, you place a dummy head with a known HRTF and two linear microphones, compensating the microphone response with the inverse HRTF, so that we factor out the head's individual HRTF, and just get the nonlinear frequency response of sound as it hits its "ears". Next, we play and record a 20Hz-20kHz linear sweep, with the speakers placed at +/-30, +/-45, +/-60, +/-90, +/-120, and +/-135 degrees in respect to the center-front of the head, and average the 6 curves. After that, we calculate the inverse curve, resulting in our target. I came up with the idea to use multiple angles, because: 1. The frequency response of our ears changes significantly depending on which angle we receive the sound; and 2. Because I've personally found that the angle of the driver in headphones can vary from within that range, despite the cups being placed at +/- 90 degrees. (I prefer the cups themselves to be angled, as well as the pads, so that the drivers are at +/-30 degree placement, just like typical speakers, as it creates extra space between the driver and the ears, and improves my ability to discern the location of sounds in music.)
But what problem am I seeking to solve with this curve, beyond the lack of a standard curve for headphones? Based on what knowledge I have, I think it makes sense to view the headphone cups behind the driver as speaker enclosures, and the cups and pads in front of the driver like a room, so that we can analyze them separately, and safely ignore the stuff in front of the driver that affects the enclosure order. Under this model, an open-back headphone would tend to be a second order (or arguably third order) enclosure, while a semi-open or closed-back headphone is a fourth order enclosure, with closed-back headphones sometimes having even more orders than that. I hypothesize that the "openness" of open-back headphones is at least partially due to the lack of backwave reflections, and that the "smaller" sound of most closed-back headphones is partially due to the innumerable backwave reflections, which I believe could be fixed by reducing them. And by offsetting drivers 30 degrees, and applying this new frequency response to both types of cans, I predict we can improve directionality and "headstage" even further. (I'm sorry for the long OP)

 

[bigshot]

Could we put the dummy head in a vice in the middle of the salt flats? How long of an extension cord are we going to need? Obviously we can't use a generator to power the equipment... maybe if the extension cord was a mile or two long... Should the extension cord be silver or copper?

 

[castleofargh]

After learning about speaker orders, in combination with being exposed to some recent threads, I came up with a new idea for developing a compensation curve for circumaural headphones: I call it the "Open Air Response" (OA).
In a flat open field, with little to no background noise, (like the California salt flats) you place two first order speakers (so that the sound travels bidirectionally) with a known frequency response, and compensate them with an inverse curve, making them linear. After that, you place a dummy head with a known HRTF and two linear microphones, compensating the microphone response with the inverse HRTF, so that we factor out the head's individual HRTF, and just get the nonlinear frequency response of sound as it hits its "ears". Next, we play and record a 20Hz-20kHz linear sweep, with the speakers placed at +/-30, +/-45, +/-60, +/-90, +/-120, and +/-135 degrees in respect to the center-front of the head, and average the 6 curves. After that, we calculate the inverse curve, resulting in our target. I came up with the idea to use multiple angles, because: 1. The frequency response of our ears changes significantly depending on which angle we receive the sound; and 2. Because I've personally found that the angle of the driver in headphones can vary from within that range, despite the cups being placed at +/- 90 degrees. (I prefer the cups themselves to be angled, as well as the pads, so that the drivers are at +/-30 degree placement, just like typical speakers, as it creates extra space between the driver and the ears, and improves my ability to discern the location of sounds in music.)
But what problem am I seeking to solve with this curve, beyond the lack of a standard curve for headphones? Based on what knowledge I have, I think it makes sense to view the headphone cups behind the driver as speaker enclosures, and the cups and pads in front of the driver like a room, so that we can analyze them separately, and safely ignore the stuff in front of the driver that affects the enclosure order. Under this model, an open-back headphone would tend to be a second order (or arguably third order) enclosure, while a semi-open or closed-back headphone is a fourth order enclosure, with closed-back headphones sometimes having even more orders than that. I hypothesize that the "openness" of open-back headphones is at least partially due to the lack of backwave reflections, and that the "smaller" sound of most closed-back headphones is partially due to the innumerable backwave reflections, which I believe could be fixed by reducing them. And by offsetting drivers 30 degrees, and applying this new frequency response to both types of cans, I predict we can improve directionality and "headstage" even further. (I'm sorry for the long OP)

while you can go in your garden and measure speakers without any walls around, I suspect that an anechoic chamber would be what you're looking for. and even less costly, you could probably just measure over a short distance and cut off the end of of the impulse to discard most if not all of the room reverb(that impulse can be created from the sweep and can then be turned back into a FR anytime if that's what you're after).
if we have to reverse the HRTF of the dummy, first why not just use a mic alone? and second, you're going to lose all the relevant cues including direction based FR. so something will have to give here in your setup.
if you measure with a sweep, you're getting the linear response as far as I know.

all in all, I'm of the opinion that we do know a great deal, or at least we have some hypotheses about what does what on headphones. the elephant that never leaves the room about positioning or just feeling flat, is always the listener's own HRTF. without knowing what you specifically need to feel flat and from a given position, a headphone can only guess or try to be designed to become the very best sound a dummy head has ever heard. I suspect that those dummy heads have been sentient being from the start and they've used us as labor force to research great headphones for them. the real world leaders and dummy heads and mice! wee know how to treat so many diseases on mice that we can't treat on humans, it's obvious that we're being used.