Starting to think about HRTF and how to measure and apply them.

[JoachimStrobel]

I came across the website of https://www.homeaudiofidelity.com/english/home/ and was intrigued that they do offer HRTF corrections. I did not find a link to an appropriate hardware or a how-to ( I could not open the supplied google drive link), but I found that topic super interesting.
Has anybody done any such measurements or corrections? What would one correct? The surround and centre speaker “timbre”? What to do for stereo? I guess the recording/mastering engineer bakes his own HRTF function into the recording and the question is, how my own HRTF could be used then.
I came across these https://de-de.sennheiser.com/in-ear-kopfhoerer-ohrhoerer-3d-ambeo-smart-headset for an unbelievable price ( English version can be googled quickly). Does anybody have any experience with those and what to do with them and how (for HRTF measurements)? I was thinking of playing REW sines or pink nose and use AudioTools on IOS to check responses. For that price it seems to be enjoyable to do.

 

[Mad_Economist]

Such measurements are at the core of both Smyth Research's Realizer and BACCH. Outside of these, there is an open source project with more limited aims.

Has anybody done any such measurements or corrections? What would one correct? The surround and centre speaker “timbre”? What to do for stereo?

Binaural microphone based corrections have typically taken two forms: attempts to match the subjective space and timbre of speaker or real sound incidence on headphones (the primary goal of the Realizer and Impulcifier), and crosstalk cancellation/advanced spatialization approaches for speakers. The latter are not my domain so I won't waste anybody's time on my conjecture.

For headphone correction, making the SPL at the eardrum in the headphone case equivalent to the SPL at the eardrum in the speaker case meaningfully improves the subjective perception of space, and for many people in many cases timbre/tone as well. It won't necessarily get you 100% of the way there, depending on your takes regarding sound localization and perception, but at minimum, making your headphones sound like a good pair of speakers to your own ears is for most people's view an improvement.

Obviously, we can't put microphones in people's eardrums - there's these pesky laws about that, and it's unergonomic - so various approaches attempt to achieve a comparable result with different positions for microphones. Past work on binaural equalization implies that a microphone set within a couple of millimetres of the ear canal, which is either fully occluded or completely unoccluded, will generally result in subjectively acceptable corrections of headphones to speakers or incident sound. Unfortunately, the Ambeo headset will not do this, as its microphones are placed markedly further out - you would still get some HRTF data which translates to the eardrum point, including head shadowing effects and interaural time differences, as well as some aspects of the pinna's response contributions, but the delta of the response at this point on a headphone and on a speaker will be different from the delta at the eardrum, so the subjective effect may be less than ideal.

That said, if you can get the Ambeo set cheap - or just want to grab or throw together a little in-ear mic on the cheap - it can be a fun project, and Jaakko's software makes it a fair bit easier if you don't wanna throw too much time at it.

I guess the recording/mastering engineer bakes his own HRTF function into the recording and the question is, how my own HRTF could be used then.

In principle, he shouldn't. As part of our brains' hearing process, we attribute a location or incidence to sounds, and this in turn impacts our perception of timbre. Imagine that you're listening to a real instrument in a room - even a 15 degree change of orientation for your head can vastly change the response at some frequencies, to say nothing of the even larger difference between your ears unless it's directly in front or behind you, but you still hear the violin, not a hundred different instruments over the course of listening.

It's been suggested that HRTF may play some role in preference variations between listeners, and if so that could influence a master from one engineer, but I'm not aware of particularly strong evidence that people want very different things when their HRTFs are controlled for. Indeed, a lot of the argument for individualized HRTF compensation on headphones is precisely that if we get the same response at two different eardrums, which would have a different response when presented with that instrument in reality, we get two different perceptions.

 

[JoachimStrobel]

Thanks a lot. I ordered that Senheiser piece for 68€ and will report back.

 

[phoenixdogfan]

Such measurements are at the core of both Smyth Research's Realizer and BACCH. Outside of these, there is an open source project with more limited aims.


Binaural microphone based corrections have typically taken two forms: attempts to match the subjective space and timbre of speaker or real sound incidence on headphones (the primary goal of the Realizer and Impulcifier), and crosstalk cancellation/advanced spatialization approaches for speakers. The latter are not my domain so I won't waste anybody's time on my conjecture.

For headphone correction, making the SPL at the eardrum in the headphone case equivalent to the SPL at the eardrum in the speaker case meaningfully improves the subjective perception of space, and for many people in many cases timbre/tone as well. It won't necessarily get you 100% of the way there, depending on your takes regarding sound localization and perception, but at minimum, making your headphones sound like a good pair of speakers to your own ears is for most people's view an improvement.

Obviously, we can't put microphones in people's eardrums - there's these pesky laws about that, and it's unergonomic - so various approaches attempt to achieve a comparable result with different positions for microphones. Past work on binaural equalization implies that a microphone set within a couple of millimetres of the ear canal, which is either fully occluded or completely unoccluded, will generally result in subjectively acceptable corrections of headphones to speakers or incident sound. Unfortunately, the Ambeo headset will not do this, as its microphones are placed markedly further out - you would still get some HRTF data which translates to the eardrum point, including head shadowing effects and interaural time differences, as well as some aspects of the pinna's response contributions, but the delta of the response at this point on a headphone and on a speaker will be different from the delta at the eardrum, so the subjective effect may be less than ideal.

That said, if you can get the Ambeo set cheap - or just want to grab or throw together a little in-ear mic on the cheap - it can be a fun project, and Jaakko's software makes it a fair bit easier if you don't wanna throw too much time at it.My Smyth


In principle, he shouldn't. As part of our brains' hearing process, we attribute a location or incidence to sounds, and this in turn impacts our perception of timbre. Imagine that you're listening to a real instrument in a room - even a 15 degree change of orientation for your head can vastly change the response at some frequencies, to say nothing of the even larger difference between your ears unless it's directly in front or behind you, but you still hear the violin, not a hundred different instruments over the course of listening.

It's been suggested that HRTF may play some role in preference variations between listeners, and if so that could influence a master from one engineer, but I'm not aware of particularly strong evidence that people want very different things when their HRTFs are controlled for. Indeed, a lot of the argument for individualized HRTF compensation on headphones is precisely that if we get the same response at two different eardrums, which would have a different response when presented with that instrument in reality, we get two different perceptions.

My Smyth A16 Realiser will be coming back from repair, and upgrade next month. I'll keep everyone posted on how well it replicates the sound of speakers in my room.