ASR Headphone Testing and BK 5128 Hats Measurement System

[solderdude]

Question.
When the mics aren't inserted deep enough how much deviation will there be in the final result.
I have seen pictures of people using in ear mics that stick out quite far. I can't imagine one can get even remotely accurate results that I suspect are essential for this to work optimally.

 

[Mad_Economist]

Question.
When the mics aren't inserted deep enough how much deviation will there be in the final result.
I have seen pictures of people using in ear mics that stick out quite far. I can't imagine one can get even remotely accurate results that I suspect are essential for this to work optimally.

Depends on the intention. Any variation in insertion depth or positioning can impact frequency response measurements both of sound sources far from the head and headphones. However, what we're primarily worried about in the Realizer/Impulcifier case would be two things: first, is any directional component of the transfer function being recorded correctly, since this will be highly significant in compensation of this sort. Per Middlebrooks et al 1989 and Hammershøi & Møller 1991, having the microphone consistently positioned within a couple of mm of the canal entrance will make its input essentially directionally independent. Second, are there interactions of the head-headphone system with the microphone present which differ from those which occur when the microphone is not present. I have seen this used as an argument for open-canal microphone designs - which, unlike occluding microphones, do not remove the canal loading element - but my own experiments in this regard have been inconclusive.

In general, unless the intention is measurement only at very low frequencies (<500hz), the microphone should be within 2-4mm of the canal entrance at most; the idyll in my opinion would be perfectly flush while being completely nonoccluding.

 

[617]

Depends on the intention. Any variation in insertion depth or positioning can impact frequency response measurements both of sound sources far from the head and headphones. However, what we're primarily worried about in the Realizer/Impulcifier case would be two things: first, is any directional component of the transfer function being recorded correctly, since this will be highly significant in compensation of this sort. Per Middlebrooks et al 1989 and Hammershøi & Møller 1991, having the microphone consistently positioned within a couple of mm of the canal entrance will make its input essentially directionally independent. Second, are there interactions of the head-headphone system with the microphone present which differ from those which occur when the microphone is not present. I have seen this used as an argument for open-canal microphone designs - which, unlike occluding microphones, do not remove the canal loading element - but my own experiments in this regard have been inconclusive.

In general, unless the intention is measurement only at very low frequencies (<500hz), the microphone should be within 2-4mm of the canal entrance at most; the idyll in my opinion would be perfectly flush while being completely nonoccluding.

Given the complexity of the inner ear, wouldn't it make more sense to simulate its resonances by capturing its shape rather than trying to stick a microphone in it?

 

[Mad_Economist]

Given the complexity of the inner ear, wouldn't it make more sense to simulate its resonances by capturing its shape rather than trying to stick a microphone in it?

I'm not aware of any efforts to simulate the inner or middle ear or how it's typically done - I'm sure it's done, but for audio purposes (outside of specific cases like bone conduction) we typically only concern ourselves with what's happening up to the eardrum.

 

[NTK]

The truly complicated thing is how the pressure waves created by the headphone drivers reach the eardrum.

They are not purely acoustic waves. Ever wonder why good bass from headphones requires good seal? Loudspeakers don't need it. Try removing your headphones slowly. When you break the seal, the low bass is gone first. As you remove the headphones further away, the mid-bass goes next, and the disappearing act progressively moves up with frequency. This is completely different from loudspeakers, where attenuation with distance (when not too far) is frequency independent.

The small size of headphone drivers make them "inefficient" acoustic radiators. The low bass propagates as "supersonic" hydrodynamic pressure waves (due to incompressibility). The mid bass (and up) are evanescent waves, where pressure attenuation is exponential with distance (instead of 1/r as normal acoustic waves). So how the "sound waves" reaches the eardrums are also fundamentally different between loudspeakers and headphones.


From An Introduction to Acoustics, which can be downloaded here: https://www.win.tue.nl/~sjoerdr/papers/boek.pdf

P.S. That's pretty much all I know about this topic. Don't make the assumption that I understand more than 0.1% of the content of the book

 

[617]

The truly complicated thing is how the pressure waves created by the headphone drivers reach the eardrum.

They are not purely acoustic waves. Ever wonder why good bass from headphones requires good seal? Loudspeakers don't need it. Try removing your headphones slowly. When you break the seal, the low bass is gone first. As you remove the headphones further away, the mid-bass goes next, and the disappearing act progressively moves up with frequency. This is completely different from loudspeakers, where attenuation with distance (when not too far) is frequency independent.

The small size of headphone drivers make them "inefficient" acoustic radiators. The low bass propagates as "supersonic" hydrodynamic pressure waves (due to incompressibility). The mid bass (and up) are evanescent waves, where pressure attenuation is exponential with distance (instead of 1/r as normal acoustic waves). So how the "sound waves" reaches the eardrums are also fundamentally different between loudspeakers and headphones.

View attachment 78847
From An Introduction to Acoustics, which can be downloaded here: https://www.win.tue.nl/~sjoerdr/papers/boek.pdf

P.S. That's pretty much all I know about this topic. Don't make the assumption that I understand more than 0.1% of the content of the book

Ok this is officially too much for me to handle. I'm going to put on Frasier and eat some cheez its.

 

[amirm]

P.S. That's pretty much all I know about this topic.

Professor Farina has a great presentation about this where he explains why rock singers often shove the microphone into their mouth to get better bass recording!

 

[Doodski]

Professor Farina has a great presentation about this where he explains why rock singers often shove the microphone into their mouth to get better bass recording!

Like this?

 

[amirm]

OK... not quite.

 

[NTK]

Found a COMSOL blog post on ear canal acoustic modeling and a headphone simulation example.
https://www.comsol.com/blogs/modeling-an-ear-canals-acoustics-to-optimize-in-ear-audio-products/
https://www.comsol.com/model/download/671031/models.aco.headphone_artificial_ear.pdf

 

[617]

Instead of torturing our technology to conform to our physiology, we should be changing our physiology to conform to our transducers. The ear canal should be a simple transmission line.

 

[pozz]

Instead aof torturing our technology to conform to our physiology, we should be changing our physiology to conform to our transducers. The ear canal should be a simple transmission line.

I have a quote from one of the authorities in audio engineering who said something like the acoustic transmission of sound will soon be a thing of the past given the developments in brain implants.

I'll try to find it.

 

[Mad_Economist]

I have a quote from one of the authorities in audio engineering who said something like the acoustic transmission of sound will soon be a thing of the past given the developments in brain implants.

(I would be interested in the quote, though)

 

[Blumlein 88]

Maybe Amir needs this instead. Would make the headphone industry boom if we had ears like this.

Omni binaural microphone setup.
https://3diosound.com/products/omni-binaural-microphone

 

[Deleted member 16543]

If there is interest I can provide a pair of ears with anatomically accurate ear canals. The canal is averaged over hundreds of measurements.
Then of course a target curve and a measuring method to get it would need to be agreed upon.
@amirm, if you are interested in looking into this, please let me know and we can talk about it.

 

[Mad_Economist]

If there is interest I can provide a pair of ears with anatomically accurate ear canals. The canal is averaged over hundreds of measurements.
Then of course a target curve and a measuring method to get it would need to be agreed upon.
@amirm, if you are interested in looking into this, please let me know and we can talk about it.

I don't know about Amir, but I'm sure as heck interested, and I'd be happy to offer comparison measurements via HATS or Etymotic Er7 probes...

Where did you get so many canal scans?

 

[Deleted member 16543]

I don't know about Amir, but I'm sure as heck interested, and I'd be happy to offer comparison measurements via HATS or Etymotic Er7 probes...

Where did you get so many canal scans?

Medical field.
Good to know there's some interest. I would be glad to contribute to this measurement endeavor, even if in such a small way.

The approach I propose is that of measuring the response of a binaural microphone at the listening position of a pair of balanced sounding speaker system.
There is a lot of good research that indicates what a balanced speaker system measures like. It's basically the inverse approach of the spinorama, where instead of measuring a source with certain optimal response and directivity and assume a standard room to get the right response at the listening position, measurements are made at the listening position and DSP is applied, altering the speaker response so that it produces the optimal sweet spot response.
Anyway, once the speaker system measures correctly at the sweet spot, I propose to put a binaural microphone there and measure the frequency response. This will be the target that a set of headphones needs to match to sound as balanced as the speakers.

There are a few details to take into account that I didn't specify, but this is basically it. It's also the same approach followed by Harman, but they did it without ear canals and focusing only on the amplitude response.

 

[Mad_Economist]

Medical field.
Good to know there's some interest. I would be glad to contribute to this measurement endeavor, even if in such a small way.

The approach I propose is that of measuring the response of a binaural microphone at the listening position of a pair of balanced sounding speaker system.
There is a lot of good research that indicates what a balanced speaker system measures like. It's basically the inverse approach of the spinorama, where instead of measuring a source with certain optimal response and directivity and assume a standard room to get the right response at the listening position, measurements are made at the listening position and DSP is applied, altering the speaker response so that it produces the optimal sweet spot response.
Anyway, once the speaker system measures correctly at the sweet spot, I propose to put a binaural microphone there and measure the frequency response. This will be the target that a set of headphones needs to match to sound as balanced as the speakers.

There are a few details to take into account that I didn't specify, but this is basically it. It's also the same approach followed by Harman, but they did it without ear canals and focusing only on the amplitude response.

For target response, this could be workable, although from a testing standpoint I would strongly suggest starting with quasianechoic free field measurements. We have a very robust idea of what the correct 0 azimuth, 0 elevation free field HRTF "should" look like, and given that a full replication of the canals to the drum is pretty novel as far as I'm aware, it's likely worth looking at how the response differs compared to conventional emulations (and, in theory, I can also test with a probe mic, as said).

A sealing earphone placed near the entrance of the canal would be an additional interesting element. I'm legitimately unsure of what an omni mic at the terminus to your canal will measure like compared to a 60318-4.

 

[Deleted member 16543]

Yes, the measurement methodology would need to be appropriate, meaning that psychoacoustics needs to be taken into account.
The target response obtained with my proposed method would be sensibly different than any other method. But that's exactly the point, I think.

 

[Moderate Dionysianism]

If there is interest I can provide a pair of ears with anatomically accurate ear canals. The canal is averaged over hundreds of measurements.

I'm achingly curious to find out whether this option was explored further, either by @amirm or @Mad_Economist or anyone else, for that matter.