• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Measuring HRTF for headphone use

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
Have you tried the equal loudness test before @pkane? I was wondering on how you did it cause I just looked back on my data and I think I may have made a mistake on my part. :0
 

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
Also, I was wondering about something. When a person goes through the equal loudness test with speakers, wouldn't the curve that we end up with be like the ear resonance curve?
Ear-resonance.jpg

Or at least the combination of all the resonances? Cause if so, my curve barley reflects that when playing tones back from my EQ'd speakers.... :0
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
Have you tried the equal loudness test before @pkane? I was wondering on how you did it cause I just looked back on my data and I think I may have made a mistake on my part. :0

I didn't use Greisinger's method. I actually did something that determined the lowest threshold of hearing using headphones. It's an equal loudness curve of sorts, but not the one you'd get from Greisinger at normal listening levels. If you are familiar with Fletcher Munson curves, it would be the one at the bottom:

media%2F255%2F255277fb-5a80-4999-b2d6-004338526cc9%2Fphp4GFAGZ.png
 
Last edited:

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
I didn't use Greisinger's method. I actually did something that determined the lowest threshold of hearing using headphones. It's an equal loudness curve of sorts, but not the one you'd get from Greisinger at normal listening levels. If you are familiar with Fletcher Munson curves, it would be the one at the bottom:

media%2F255%2F255277fb-5a80-4999-b2d6-004338526cc9%2Fphp4GFAGZ.png
Oh... But we're you able to produce or get a curve like I mentioned in my last post?
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
Oh... But we're you able to produce or get a curve like I mentioned in my last post?

It's different, because it reflects a combination of HRTF, my hearing threshold, and HE560 headphone frequency response. It's also the inverse of the curve you posted, since a resonant frequency (a high point on your chart) would produce a lower threshold (lower point) on my chart.
 

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
Oh. Hmmm... So you have three curves you add all together to get tge best sound out of your headphones? Seems like you're results put mine to shame cause even with the curve I got through the equal loudness test, it didn't sound as good as I might hope... And I think it's because I have no way of getting the raw headphone response to calibrate them.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
Oh. Hmmm... So you have three curves you add all together to get tge best sound out of your headphones? Seems like you're results put mine to shame cause even with the curve I got through the equal loudness test, it didn't sound as good as I might hope... And I think it's because I have no way of getting the raw headphone response to calibrate them.

Yes, but that's only something I measured. I didn't try to adjust to this curve, as it was fairly low resolution and probably not extremely accurate. I EQ'ed using the curve measured by the in-ear microphones, which is distinctly different.
 

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
Hmmm... But what I'm not understanding is to why getting or using the Fletcher Munson curve will help you... :/
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I'll try to un-confuse you.

The HRTF curve is ONLY relevant when measuring speakers in front of the HATS in an anechoic room for that specific HATS.
The ear resonance needs to be compensated when measuring.
The reason is the construction of the HATS mimics what is measured at ear drum level when sound is coming from in front of it.

Microphones are linear. They (should) give the same output voltage at all frequencies at the same sound pressure.
The ear is non-linear and also SPL dependent.
That last bit is where the Fletcher-Munson part becomes relevant when you want to relate perceived loudness (Phon) to a measured SPL (in dB or Pascal).

The Fletcher-Munson curves are relevant when a recording is mixed using speakers and ears at a certain SPL under the condition that the speakers are 'flat' and the mixing engineer knows how real instruments/voices sound.
When the sound is 'mixed' at say 80dB average the correct (perceived) tonal balance will be heard when reproduced at the same SPL.
At a lower SPL bass will sound 'lower' in level, at a higher SPL bass will sound relatively too loud.
I made an illustration which is part of my explanation HERE
f-m-overlayed.png

This is why you should do the Griesinger bit at the correct SPL level (SPL meters are cheap on e-bay and close enough to be correct-ish)

When measuring speakers with a HATS the ear resonance is important as it has to be 'undone' again to get the SPL at the speaker in front of the HATS in an anechoic room.
When measuring speakers in front of the head+mic in an anechoic room with the mics Pkane is using the earcanal-eardrum correction must not be applied as there is none so no correction is needed. But the Pinna, Torso and Concha (1,2,3 and 4) need to be applied when you want to know what the (approximate, not exact because explained earlier).

When measuring speakers with a wand mic in an anechoic room no compensation is needed at all. You measure what is there. But you don't percieve it as 'flat' depending on the SPL (F-M). The microphone does not care, it is linear, the ears are not.

Measuring headphones requires a different correction when using HATS and In-ear mics when sounds come from a planar field on the side of the measuring microphone. When one also wants to correlate to perceived sound SPL comes into play (F-M) and because the 'body' also inputs to the brain which is not there with headphones. A bees nest.. stay away from this unless you calibrate each step when you want to achieve relevant and accurate plots. A home made plot MAY have some relevance to perceived sound but is NOT something you can rely on.


Griesinger method:
Determine equal loudness with a speaker in front of you at a specific SPL (F-M, hearing damage and speaker FR is compensated to YOUR HRTF)
Once this is determined that becomes a REFERENCE when you want to achieve a 'speaker in front of you' sound when using headphones.
But is SPL dependent and ear dependent and NOT a reference to actual SPL at all.

Once you have that 'reference' of how YOU perceive 'flat' as in equal loudness of a speaker in front of you then you ca proceed with the headphone measurement (using Griesingers program with noise bands, not tones)
You repeat the equal loudness thing but with headphones.
When you have done that you have the same tonal balance reference of a speaker directly in front of you.
BUT you want the speaker in room experience. To reach that you have to subtract the 2 obtained 'curves' you set in the EQ.
This way (when done properly at the same perceived loudness) you get same tonal balance of a 'flat speaker' in room in front of you.

It is utterly pointless to 'confirm' this with measurements as these are NOT accurate.
The 'difference' in EQ settings that comes out of this procedure is YOUR HRTF. This does not look like the HRTF plot from the HATS at all but is the difference between the sound coming from the front and from that specific headphone, not all headphones, not your 'general' or average HRTF.
Just that 'correction' for 1 specific headphone.
It thus is rather irrelevant to other headphones and cannot be used as a 'universal' correction for your head. This is because all headphones need different compensations.
That is why Griesinger mentions the same procedure must be repeated with each (different) headphone and he has 'memory banks' to store a few.

Moral of the story... Griesingers method is great for compensating headphones. When you want speaker in room effect the speaker thing must be done as well.
Don't try to confirm your found FR with measurements as this is not accurate for the reasons I mentioned in previous posts.

Either go the measurement route with calibrated equipment (impossible to do properly within 1 or 2 dB at home) or do it the Griesinger way.
Forget about attempting to combine both methods unless everything is calibrated.
Trying to find/make a microphone calibration based on Griesingers method would only be somewhat relevant when each headphone gets its own 'calibration'. The EQ is already found so no need to base it on it.
 
Last edited:

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
I'll try to un-confuse you.

The HRTF curve is ONLY relevant when measuring speakers in front of the HATS in an anechoic room for that specific HATS.
The ear resonance needs to be compensated when measuring.
The reason is the construction of the HATS mimics what is measured at ear drum level when sound is coming from in front of it.

Microphones are linear. They (should) give the same output voltage at all frequencies at the same sound pressure.
The ear is non-linear and also SPL dependent.
That last bit is where the Fletcher-Munson part becomes relevant when you want to relate perceived loudness (Phon) to a measured SPL (in dB or Pascal).

The Fletcher-Munson curves are relevant when a recording is mixed using speakers and ears at a certain SPL under the condition that the speakers are 'flat' and the mixing engineer knows how real instruments/voices sound.
When the sound is 'mixed' at say 80dB average the correct (perceived) tonal balance will be heard when reproduced at the same SPL.
At a lower SPL bass will sound 'lower' in level, at a higher SPL bass will sound relatively too loud.
I made an illustration which is part of my explanation HERE
f-m-overlayed.png

This is why you should do the Griesinger bit at the correct SPL level (SPL meters are cheap on e-bay and close enough to be correct-ish)

When measuring speakers with a HATS the ear resonance is important as it has to be 'undone' again to get the SPL at the speaker in front of the HATS in an anechoic room.
When measuring speakers in front of the head+mic in an anechoic room with the mics Pkane is using the earcanal-eardrum correction must not be applied as there is none so no correction is needed. But the Pinna, Torso and Concha (1,2,3 and 4) need to be applied when you want to know what the (approximate, not exact because explained earlier).

When measuring speakers with a wand mic in an anechoic room no compensation is needed at all. You measure what is there. But you don't percieve it as 'flat' depending on the SPL (F-M). The microphone does not care, it is linear, the ears are not.

Measuring headphones requires a different correction when using HATS and In-ear mics when sounds come from a planar field on the side of the measuring microphone. When one also wants to correlate to perceived sound SPL comes into play (F-M) and because the 'body' also inputs to the brain which is not there with headphones. A bees nest.. stay away from this unless you calibrate each step when you want to achieve relevant and accurate plots. A home made plot MAY have some relevance to perceived sound but is NOT something you can rely on.


Griesinger method:
Determine equal loudness with a speaker in front of you at a specific SPL (F-M, hearing damage and speaker FR is compensated to YOUR HRTF)
Once this is determined that becomes a REFERENCE when you want to achieve a 'speaker in front of you' sound when using headphones.
But is SPL dependent and ear dependent and NOT a reference to actual SPL at all.

Once you have that 'reference' of how YOU perceive 'flat' as in equal loudness of a speaker in front of you then you ca proceed with the headphone measurement (using Griesingers program with noise bands, not tones)
You repeat the equal loudness thing but with headphones.
When you have done that you have the same tonal balance reference of a speaker directly in front of you.
BUT you want the speaker in room experience. To reach that you have to subtract the 2 obtained 'curves' you set in the EQ.
This way (when done properly at the same perceived loudness) you get same tonal balance of a 'flat speaker' in room in front of you.

It is utterly pointless to 'confirm' this with measurements as these are NOT accurate.
The 'difference' in EQ settings that comes out of this procedure is YOUR HRTF. This does not look like the HRTF plot from the HATS at all but is the difference between the sound coming from the front and from that specific headphone, not all headphones, not your 'general' or average HRTF.
Just that 'correction' for 1 specific headphone.
It thus is rather irrelevant to other headphones and cannot be used as a 'universal' correction for your head. This is because all headphones need different compensations.
That is why Griesinger mentions the same procedure must be repeated with each (different) headphone and he has 'memory banks' to store a few.

Moral of the story... Griesingers method is great for compensating headphones. When you want speaker in room effect the speaker thing must be done as well.
Don't try to confirm your found FR with measurements as this is not accurate for the reasons I mentioned in previous posts.

Either go the measurement route with calibrated equipment (impossible to do properly within 1 or 2 dB at home) or do it the Griesinger way.
Forget about attempting to combine both methods unless everything is calibrated.
Trying to find/make a microphone calibration based on Griesingers method would only be somewhat relevant when each headphone gets its own 'calibration'. The EQ is already found so no need to base it on it.

Ahh!! I bet that's why I couldn't get it right... I'll make sure my speakers are loud enough next time. I do have an SPL app on my phone and I can use my Dayton imm-6 with it and the calibration file too. I may have to recalibrate the meter though for the quietest environment, right?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
It depends on what you plan to do.... now I am confused.
can you walk me through the consecutive steps you want to do in order to achieve what exactly ?
 

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
It depends on what you plan to do.... now I am confused.
can you walk me through the consecutive steps you want to do in order to achieve what exactly ?

Well, what I'm trying to achieve is finding my frontal HRTF doing the method Griesinger came up with in his video. I didn't regard the Fletcher Munson curve cause I did the speaker equal loudness test at a fairly low volume (that I don't know what SPL level it was) which resulted in results that I liked once I subtract against my headphone equal loudness curve. I'll make sure next time to measure the SPL on my speakers first and adjust accordingly to the highest listening level I enjoy music in and see if my results improve. :0
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
You don't need to check your speakers for linearity.
This is already included in the Griesinger test including possible hearing damage.

For the Griesinger test it is important to do the speaker and headphone test at the same level which can be achieved by playing both at the same time and listen to the headphone and speakers alternatively (putting cans on and off quickly and adjusting the volume)

I presume you want to use your speakers. In that case set them up outside and test there otherwise room resonances come into play.
When you want this done indoors you need a small point source speaker in close proximity of you and away from walls and other surfaces.
So on a stand and close by.
NO need to know the exact SPL, just use the one you normally listen to (use music to determine this)
NO need to know the FR response of the speaker either, it too is already incorporated in the test and the resulting EQ.


You won't be able to get YOUR specific and unique HRTF in an absolute sense.
You can only get A 'correction' for one specific headphone at a set SPL.
That correction you then get will differ from another one taken at a higher or lower SPL because of F-M.

It cannot be applied to other headphones so 'knowing' your HRTF is rather pointless as you can't do anything with it.
Each headphone needs its own HRTF with the Griesinger method as well so your found HRTF with headphone A can not be applied to headphone B.
Both A and B need their own 'correction'.
 

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
I see. I thought what he did in the beginning when he mentioned that he had a calibration file for his speakers. If I don't need to worry about that, then I can just go ahead and use my small Logitech speakers I have at home. :)
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I believe pkane just was experimenting and tried a few methods to arrive at his HRTF (a single HRTF does not exist for all different headphones).
To rule out F-M and his ear condition he was trying to see if he could sort of mimic Griesingers method but with measurements using 2 different mics and under different circumstances to find 1 correction he could apply to all headphones. A bit like the HRTF from a HATS but without the ear canal resonance. This stands or falls with references where some (in this case) would be relying on assumptions that the references would be accurate.
A room correction slope is also what he was after, not 'flat' in anechoic conditions. A more elaborate and difficult method he wanted to use to arrive at a more scientifically based correction for all headphones at once (like a HATS).

It is much easier and less prone to deviations to simply use the Griesinger method IF you want to arrive at 'flat speaker in room' correction for headphones. It would have to be followed to the letter.
These Logitech speakers need to be close to you (like in the video) and must be able to reproduce both the lower frequencies and highest frequencies (MUST be noise bands in this case) with the EQ the program is going to apply to get equal loudness of all frequencies at the desired levels. They should be single speaker (point source) but you could try it with 2-way if that is what you have.
That's why Schrödinger used an extensive (stereo) speaker. The small $15.- ones he showed in his video for certain won't be reproducing the lower frequencies at the required SPL and could skew the results.

I can't speak for pkane but my assumption based on earlier remarks is that he would add a 'room curve' to the found EQ this way as one would be referencing to a speaker flat in anechoic conditions (that is what Griesingers method actually attempts) but don't think this is needed as the ears are also already 'corrected' in this test.
Basically:
errors in the speaker are already 'corrected' (that's why you need to do the speaker test first)
The room is already corrected (small speakers close-by and thus room effects are much less (if at all) relevant
Abbreviations in your hearing are already 'corrected' both in speaker and headphone mode when performed at a comfortable listening level you prefer to listen at.
The EQ obtained that way (in the speaker reference test) thus yields an EQ with speaker + ear (but coming in front of you).
Then the same method is used but speaker in front is replaced with a (specific) headphone. (acc. to Griesinger the apple inear is the best of all)
The EQ obtained that way yields an EQ including FR of the headphone + ear (but from the side of your head).

So basically you have found 2 different EQ's this way.
The interesting part of Griesingers method here is that the 'trick' lies in the switching frequency noise bands which the ear, used as a measuring device now, is assumed to have corrected the FR issues of speaker in anechoic conditions (speakers close by) and the headphone in the EQ.
This way the difference that remains between the EQ's thus is: sound coming from in front of you AND sound coming from a headphone.
The whole idea is to 'correct' the sound coming from the side to sound coming from in front of you. That is in tonal balance.
This is achieved by substracting the 2 EQ's which then is 'added' to the headphone correction file.

Trying to plot both the speaker and headphone correction and then subtracting them won't yield the HRTF so is pointless (which is what I think you are after as well).
The reason for this is that both obtained EQ's include more than 1 variable. When the speaker would be absolutely 'flat' and the headphone would also be absolutely 'flat' that could be done this way. BUT both the speaker and headphone FR are both also corrected so the difference between both EQ's consists of FR correction for both ears + speaker FR + headphone FR. Measuring speaker FR is possible with a reasonable accuracy, measuring headphones is prone to have errors of many dB's. When that would be done perfectly the difference file would yield the difference between both ears + HRTF (for that specific headphone only).
So don't bother 'checking' your found HRTF afterwards using 'science' as there are too many unknowns which cannot be pulled apart later on as the measurements are not referenced. This means you would just see a couple of measurements which are all highly suspect.

Just repeat the Griesinger test as in the video using the same program and method and then enjoy the found sound.
When you don't like the tonal balance this way just add a little 'tone control' to suit your taste and forget trying to reason why nor to 'check' all steps using 'scientifically based' steps. Reason: proper calibration of measurement gear. Assumptions that what one sees on the screen obtained through mics is merely an illusion in this case and should not be used to obtain the same sonic end result. It simply won't.

Do the test (for each headphone you own/use) and start enjoying music without having to wonder about tonal balance.

All that remains now is the accuracy of the headphone. Which is flawed with resonances, peaks and dips and coloration which aren't corrected.

Don't think that all headphones now sound the same. They won't, they will all still sound quite different but their tonal balance will be closer to each other for sure. Others can not use your found EQ either because the hearing (which is included) won't be the same.

Best results will be with the highest quality headphone that also needs the least amount of 'gradual, overall correction' simply because sharp nulls and peaks are not 'corrected' this way and will still color the sound.
The headphones with the best 'performance' (without correction) thus will give the best sound even in this case.

Measuring HRTF for headphone use (thread title) thus is not really feasible for a number of reasons.
Finding HRTF-compensated EQ using the ears only (for each headphone) is more do-able but found EQ cannot be used to show your HRTF.
 
Last edited:

Shikamon

Member
Joined
Sep 3, 2018
Messages
80
Likes
7
I believe pkane just was experimenting and tried a few methods to arrive at his HRTF (a single HRTF does not exist for all different headphones).
To rule out F-M and his ear condition he was trying to see if he could sort of mimic Griesingers method but with measurements using 2 different mics and under different circumstances to find 1 correction he could apply to all headphones. A bit like the HRTF from a HATS but without the ear canal resonance. This stands or falls with references where some (in this case) would be relying on assumptions that the references would be accurate.
A room correction slope is also what he was after, not 'flat' in anechoic conditions. A more elaborate and difficult method he wanted to use to arrive at a more scientifically based correction for all headphones at once (like a HATS).

It is much easier and less prone to deviations to simply use the Griesinger method IF you want to arrive at 'flat speaker in room' correction for headphones. It would have to be followed to the letter.
These Logitech speakers need to be close to you (like in the video) and must be able to reproduce both the lower frequencies and highest frequencies (MUST be noise bands in this case) with the EQ the program is going to apply to get equal loudness of all frequencies at the desired levels. They should be single speaker (point source) but you could try it with 2-way if that is what you have.
That's why Schrödinger used an extensive (stereo) speaker. The small $15.- ones he showed in his video for certain won't be reproducing the lower frequencies at the required SPL and could skew the results.

I can't speak for pkane but my assumption based on earlier remarks is that he would add a 'room curve' to the found EQ this way as one would be referencing to a speaker flat in anechoic conditions (that is what Griesingers method actually attempts) but don't think this is needed as the ears are also already 'corrected' in this test.
Basically:
errors in the speaker are already 'corrected' (that's why you need to do the speaker test first)
The room is already corrected (small speakers close-by and thus room effects are much less (if at all) relevant
Abbreviations in your hearing are already 'corrected' both in speaker and headphone mode when performed at a comfortable listening level you prefer to listen at.
The EQ obtained that way (in the speaker reference test) thus yields an EQ with speaker + ear (but coming in front of you).
Then the same method is used but speaker in front is replaced with a (specific) headphone. (acc. to Griesinger the apple inear is the best of all)
The EQ obtained that way yields an EQ including FR of the headphone + ear (but from the side of your head).

So basically you have found 2 different EQ's this way.
The interesting part of Griesingers method here is that the 'trick' lies in the switching frequency noise bands which the ear, used as a measuring device now, is assumed to have corrected the FR issues of speaker in anechoic conditions (speakers close by) and the headphone in the EQ.
This way the difference that remains between the EQ's thus is: sound coming from in front of you AND sound coming from a headphone.
The whole idea is to 'correct' the sound coming from the side to sound coming from in front of you. That is in tonal balance.
This is achieved by substracting the 2 EQ's which then is 'added' to the headphone correction file.

Trying to plot both the speaker and headphone correction and then subtracting them won't yield the HRTF so is pointless (which is what I think you are after as well).
The reason for this is that both obtained EQ's include more than 1 variable. When the speaker would be absolutely 'flat' and the headphone would also be absolutely 'flat' that could be done this way. BUT both the speaker and headphone FR are both also corrected so the difference between both EQ's consists of FR correction for both ears + speaker FR + headphone FR. Measuring speaker FR is possible with a reasonable accuracy, measuring headphones is prone to have errors of many dB's. When that would be done perfectly the difference file would yield the difference between both ears + HRTF (for that specific headphone only).
So don't bother 'checking' your found HRTF afterwards using 'science' as there are too many unknowns which cannot be pulled apart later on as the measurements are not referenced. This means you would just see a couple of measurements which are all highly suspect.

Just repeat the Griesinger test as in the video using the same program and method and then enjoy the found sound.
When you don't like the tonal balance this way just add a little 'tone control' to suit your taste and forget trying to reason why nor to 'check' all steps using 'scientifically based' steps. Reason: proper calibration of measurement gear. Assumptions that what one sees on the screen obtained through mics is merely an illusion in this case and should not be used to obtain the same sonic end result. It simply won't.

Do the test (for each headphone you own/use) and start enjoying music without having to wonder about tonal balance.

All that remains now is the accuracy of the headphone. Which is flawed with resonances, peaks and dips and coloration which aren't corrected.

Don't think that all headphones now sound the same. They won't, they will all still sound quite different but their tonal balance will be closer to each other for sure. Others can not use your found EQ either because the hearing (which is included) won't be the same.

Best results will be with the highest quality headphone that also needs the least amount of 'gradual, overall correction' simply because sharp nulls and peaks are not 'corrected' this way and will still color the sound.
The headphones with the best 'performance' (without correction) thus will give the best sound even in this case.

Measuring HRTF for headphone use (thread title) thus is not really feasible for a number of reasons.
Finding HRTF-compensated EQ using the ears only (for each headphone) is more do-able but found EQ cannot be used to show your HRTF.

I see now. I just did like you said and I made sure to set my headphones and speakers at a volume that I listen to music (or the maximum I would listen to) and it sounds great! I saw a great change in my headphone equal loudness curve cause I did it at a fairly low volume. Once I got the curves, I subtracted the difference between the two, giving me the curve that only works with my headphone of choice. I believe this is what I was looking for. It was starting dead at my face the whole time. xD
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Job well done it seems and enjoy the music !
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,632
Likes
10,205
Location
North-East
I believe pkane just was experimenting and tried a few methods to arrive at his HRTF (a single HRTF does not exist for all different headphones).
[....]
Measuring HRTF for headphone use (thread title) thus is not really feasible for a number of reasons.
Finding HRTF-compensated EQ using the ears only (for each headphone) is more do-able but found EQ cannot be used to show your HRTF.

Hi Solderdude,

This thread has gone on for a while, and I'm sure I'm responsible for some of the confusion. What I was attempting to do from the beginning was a bit of research on how to adjust my stereo equipment for best sounding, most realistic playback using measurements. I learn by doing, so reading research papers and recommendations from others is just an incentive for me to do my own experiments :)

I was never after a single, general HRTF curve or even a single correction that would work on more than one set of headphones or one set of ears. In fact, the goal was the opposite: to find a reproducible method for measuring all the frequency distortions in my playback system (including my head, my ears, hearing, etc.) and to then try to correct for them.

I believe I've found a method that works well for me using the in-ear headphones. In the process, I've discovered a few other methods to achieve similar results. From Greisinger to F-M like curve measurements, I now have a much better understanding of what effect each of the FR distortions has on music reproduction.

HRTF measurement, specifically, was also part of a much longer term project for me. I'd like to be able to correct for sound directionality based on head position, without spending thousands of dollars on one of the available systems that do head tracking. But this is far off into the future. I consider this thread a success, and I apologize to @Shikamon or anyone else that I may have confused in the process. :)
 
Top Bottom