The No.1 Website for Pro Audio
A better binaural microphone
Old 4th May 2017
Lives for gear
sax512's Avatar

A better binaural microphone

This thread is intended to share my personal take on binaural microphones, their typical design approximations, the assumptions underlying them and my solutions to overcome them.
In the end, I hope this will help DIYers to make better binaural recordings.

Binaural recordings, simply put, are aimed at reproducing at the listener's eardrums, during playback, the same sound pressure that would hit their eardrums if their head was in the place of the microphone during the recording.
This way the recording achieves the highest realism possible.

The traditional way to go about this has been, so far, to put two microphone capsules right at the ear canal openings or inside shortened approximations of the ear canals. Both DIY and professional microphones use this approach (when they don't do away with the ears altogether, which is when the microphone becomes more of a variation of a spaced pair rather than a binaural microphone).
The assumption with the traditional approach is that capturing the pressure at the microphone's ear canal opening is enough to be able to reproduce the same pressure at the listener's canal opening during playback.
Unfortunately, the mere fact of putting a hard surface (the capsule) where there should be an opening (the canal) is enough to drastically modify the characteristics of the pressure wave at the capsule's location, as opposed to when the pressure wave is free to enter the canal opening.
In other words, the capsule doesn't capture the pressure that would be at its location if the capsule itself wasn't there.
Mathematically, this boundary condition is equal to forcing the wave velocity to be identically zero at the canal opening, which would not happen with an obstruction-free canal (real listening condition).

The necessary condition (although not sufficient. More on this below..) to replicate at the listener's eardrums the correct sound pressure during playback is to capture that pressure, unmodified, during the recording phase.
This means that the geometry of the outer ear (both pinna AND ear canal) has to be preserved, and that the microphone capsule needs to be placed at the canal's end, where the eardrum would be.

Some people try to achieve that by inserting small microphone probes in their ear canals, as close to their eardrums as possible, and sit there and hit record.
There are a few issues with this method.
First is the risk to hurt your eardrums if you're not careful.
Second, the fact that you have to sit there, still, for the whole duration of the recording (and not make any sound).
Third, the fact that a small probe is not so small if compared to the dimensions of the ear canal. Therefore, it shapes the pressure wave by the mere fact of being there, just like with the capsule at the canal's entrance approach, even though not quite as much.

The solution to these problems is to use anatomically accurate replicas of the whole outer ear (pinna + canal) in a binaural microphone, and place the microphone capsules at the end of the ear canals.
This way the necessary condition above mentioned is satisfied.
Unfortunately for me I had to create such replicas myself from scratch, since they were nowhere to be found for sale when I started investigating these issues.
So I embarked on a project that took me about two years, to come up with 3D printable models that I finally was able to use with my DIY binaural microphone.
These models feature ear canals that are averaged over several hundreds of real impression scans.
This is, I believe, the best choice possible to guarantee that the recordings will work for the vast majority of listeners.

The 3D printed ears can be seen and purchased here:

Pictures are linked at the bottom. The cylinder at the end of the ear canal is the microphone capsule's housing.

As far as the sufficient condition (provided that the necessary one is satisfied), what needs to be applied to the raw recording in the mixing phase is an equalization that takes into account the fact that the recorded signal will be processed again by a second set of ears (the listener's) during playback.
This equalization is already applied inside the microphone in professional models such as the Neumann KU100, with dedicated circuitry. That is not the case with DIY microphones, which are not as plug-and-play as professional ones.
However, if one is handy enough to build a DIY binaural microphone, the equalization process should be relatively simple for them to perform.

The equalization can be found as follows. While I am aware that others may use different variations of it, this procedure derives directly from the original scope of reproducing at the listener's eardrums the same sound pressure that the microphone capsules recorded.
1. Place the binaural microphone in your sweet spot.
2. Play a white noise signal from the left speaker and analyze the frequency content recorded by the left capsule.
3. Reverse it and apply it to the raw left signal recording (this can be done with Logic MatchEQ, for example).
4. Do the same for the right channel.

This procedure should be executed carefully, considering that the recorded signal could be spoiled by the room resonances.
What I do to mitigate this is I use low levels of white noise in step 2, and I open the door to the room. Also, my room is treated.
Using this procedure I have achieved consistent and repeatable realistic recordings, which is the main reason why I chose the binaural recording technique to begin with.

I strongly advise binaural microphone DIYers to at least use plastic tubes about 2.5 cm long as ear canals, in addition to the pinnae replicas, place the capsules at the canals' ends and follow the equalization procedure above.
This simple modification will get you a little closer to true 3D soundstage and realistic sound recordings, while for best results my 3D printed ears would be ideal.
Attached Thumbnails
A better binaural microphone-inside-l.jpg   A better binaural microphone-diy-binaural-microphone.jpg   A better binaural microphone-ear-detail-2.jpg   A better binaural microphone-ear-detail.jpg  
Old 4th May 2017
Lives for gear
jimjazzdad's Avatar
How does your approach differ from the Neumann KU 100? (other than cost). What microphones do you use at the end of the canal and what is the diaphragm diameter with reference to a typical human ear drum? Have you done an ABX comparison of your kunstkopf to the Neumann? Curious minds want to know...

Old 4th May 2017
Lives for gear
Plush's Avatar
The German government spent millions of Deutschmarks in research, trial, and confirmation to invent the modern binaural head. Bell Labs also spent millions of USD to refine and test binaural heads. Philips in the Netherlands invested heavily.

So today, we can rely on this former research and development work to inform us.

The ears on the Neumann head are an averaged shape developed after measuring 1200 sets of human ears.
Old 4th May 2017
Lives for gear
sax512's Avatar

The approach with my ear models is to record the pressure at the eardrums rather than at the canal's entrance. Measuring the pressure at the canal entrance is not only sub-optimal, but also impossible for the reasons explained in my initial post.
As far as the diaphragm diameter, what matters the most is the fact that the modeled canal ends with an area that has the same dimensions of an average eardrum. This leaves the geometry of the outer ear (pinna AND canal) unchanged, and so it doesn't modify the pressure wave.
As long as the capsule is less than 6mm in diameter and it fits in the cylinder housing, it is up to personal preference which one to use. Personally, I use Primo capsules with high SPL handling.

Physics tells us that adding a boundary condition like the hard surface of a capsule at the canal's opening changes the pressure wave at that location.
There is no doubt that a lot of good research has been done about the ear pinna and its characteristics.
Unfortunately the same can't be said about the whole ear (pinna + canal). So much so that I had to create these models to address this issue.
Old 23rd June 2017
Lives for gear
sax512's Avatar

Two recordings made with my DIY binaural microphone.

A few notes:
I provided the singer with a track of the whole performance.
Individual pieces cuts and video edits were done by her.
She asked for a little warmer sound, so I added some extra equalization.
A little clarity loss was the price to pay, on my opinion.
Other than that, all that I used was the raw sound and the binaural microphone's own equalization, calculated as per post #1 . No extra plugins of any kind have been used (except a limiter that comes into play only during the applause).

Capsules: matched pair Primo EM194
Naiant PFA phantom power adaptors
Mogami microphone cables
Focusrite Saffire PRO 24 audio interface
DAW: Logic pro 8
Microphone's placement: ~20-23 ft, in the center aisle, right behind the audience seats
Old 23rd June 2017
Deleted User
Neumannäs first dummy head microphone the KU80 included the earchannels in the recorded sound, the KM83 membranes were mounted inside the dummy head. When Neumann came with a succesor model, the KU81, the membranes were placed in the pinnae. I assume they had a good reason for that. Playback of recordings made with the KU80 resulted in doubling the the combfilter effect earchannel, which seems not a good idea in my opinion. B&K also uses microphones placed in the pinnae and not behind the earchannels.
Old 23rd June 2017
Lives for gear
sax512's Avatar

Thank you for the info, Adorno. I didn't know Neumann had started with a replica of the ear canals. I found this page that has a couple pictures of it.
It looks like the capsules are a little offset compared to where the eardrums would naturally be, but what I gathered is that they started with the right idea.
I do not know why they transitioned to capsules mounted at the canal's entrance later on.
My first guess would be that equalizing just the pinna is much more simple than equalizing pinna and ear canal (full ear), if one were to use dedicated circuitry like the Neumanns do.
So I guess that the natural frequency shape applied by the full ear was too much to handle back in the '70s.

All I can tell you is that it doesn't seem to be a problem with DSP, and I'm just using Logic Pro MatchEQ. There are a lot more sophisticated and accurate means to do it, but MatchEQ seems to do its job egregiously for now.
When I went from capsules at the canal's entrance to adding the ear canal, the equalization required changed quite drastically. Again, not a big problem if done with DSP, as the procedure to find the equalization is the same in both cases. Just the results are different.
The comb filtering effect you mentioned is the extra EQ necessary to account for the ear canal addition. I don't seem to have any problem with compensating for it. Incidentally, there is a secondary added benefit to the ear canal. And that is an increase in the S/N ratio.
But the main benefit I did gain when adding ear canals was a lot more accuracy in localization and tone realism.

I would love to see Neumann go back to its original idea and model ear canals again. I'm sure with today's technology they would make a killer microphone. My guess is that I would probably hate to pay for it, though..
Old 23rd June 2017
Gear Addict
CharlesCola's Avatar

Originally Posted by Adorno View Post
Neumannäs first dummy head microphone the KU80 included the earchannels in the recorded sound, the KM83 membranes were mounted inside the dummy head. When Neumann came with a succesor model, the KU81, the membranes were placed in the pinnae. I assume they had a good reason for that. Playback of recordings made with the KU80 resulted in doubling the the combfilter effect earchannel, which seems not a good idea in my opinion. B&K also uses microphones placed in the pinnae and not behind the earchannels.
I agree exactly with Adorno and this is the point I was trying to make in my previous post on your other thread.

Your recordings don't demonstrate the true power of binaural i.e. 3d localisation so are not really a good example. They do however highlight the extensive phase distortions introduced by your signal path to the capsule and method of complex eq (necessary due to placement of capsule behind pseudo ear canal in your design) these distortions are very apparent in the recordings and I'm sorry to say the resulting quality (or lack thereof) demonstrates this point.

Please check this youtube example where the capsule is at the ear opening:

The recording is much cleaner and has amazing pin point localisation and 3d effect. Due to the capsules location very near the opening there is minimum phase distortion and filtering taking place so the result is clean and easy to deal with, only requiring slight eq cut to account for eq bump caused by pinna. More importantly it is a pleasure to listen to.

Personally I would drop your method of ear canal modelling and capsule placement just as Neumann and most other binaural music mic manufacturers have. If you are capturing complex recordings of exactly what the eardrum receives (including all the comb filtering and phase distortions) for scientific research then there much more accurate models available in academic establishments - the resulting recordings are of no real use for anything other than complex scientific research. Recording listenable music that is a pleasure to listen to is a quite different (to some degree) field and requires a different approach - adding a complex layer of filtering and phase distortion caused by burying mic in what is essentially a curved tube is only going to destroy the localisation, phase coherence and other attributes of a recording which make it a joy to listen to. The main reason for this is that it is happening twice (once for the recording then once again when listening). The end result is even worse after "correction" which further shreds the signal - no matter what you do to that signal it will always be severely compromised by the roller-coaster ride the signal has taken before it reaches the capsule.

EDIT: Also you listed EM194 as the capsules used - as far as I can see there is no such capsule model from Primo - the closest to that model number is EM184 - if you used EM184 this is a cardioid capsule - binaural mics generally use omni - using a cardioid capsule would severely compromise the result (further in addition to mounting the capsule at the end of a curved tube).

Last edited by CharlesCola; 23rd June 2017 at 01:01 PM.. Reason: added question about capsule
Old 23rd June 2017
Lives for gear
sax512's Avatar

Well, Charles. If you really want to compare recordings you should do that is the same conditions.
I'm sure you realize a recording studio session is a different situation than a rehearsal room with audience. Also, the 20+ ft mic location might not be the best to capture the spread of a sound source, but that was a requirement of the client, so non negotiable.
Still, I think my recordings managed to convey the 3D sound of the space very well, but I guess that's where we disagree.

To me, the recording you posted sounds great, but in a way, bigger than life. Some people like that, and I think it's because they expect sound from a playback system to sound a certain way. But I strive for absolute realism. It's a different goal.
A word of advice from someone that's been playing classical guitar for 20+ years and started this whole venture to be able to record it realistically: classical guitars don't sound like that.

I'm sorry you seem to detest my recording. Hopefully you'll be part of the minority.

P.S. No mistake in the capsule's model name. You can contact Primo to order them directly. I strongly advise using them, as they are balanced.
Old 22nd November 2017
Lives for gear
sax512's Avatar

Old 23rd November 2019
Lives for gear
sax512's Avatar

I finally took some time to gather my thoughts in a somewhat technical compendium on everything related to binaural recordings and tried to write it down in a concise and simple way. I've been putting it off for a while, but it really should have been done when I came out with my ear replicas, especially in light of the push back I've seen by some 'traditionalists' and as a strong proof supporting the reasoning behind the need for ear canal replicas, among other considerations about binaural microphones in general.
Plus, I dreaded doing it because of the less than ideal word processor to write formulas in a forum. Hence, the very compact form of it. Better late than never.

Binaural Theory
Old 20th April 2020
Lives for gear
sax512's Avatar

B&K gets it. They are using anatomically accurate ear canals too, in their new HATS microphones.
Old 20th April 2020
Lives for gear
sax512's Avatar

G.R.A.S. follows suit..
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 103 views: 8890
Avatar for Alex Niedt
Alex Niedt 13th December 2010
replies: 4404 views: 1158897
Avatar for Dallon426
Dallon426 21 minutes ago
replies: 50 views: 12396
Avatar for Earcatcher
Earcatcher 7th September 2015
replies: 30 views: 15331
Avatar for sax512
sax512 21st April 2020
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.

Slide to join now Processing…
🖨️ Show Printable Version
✉️ Email this Page
🔍 Search thread
🎙️ View mentioned gear
Forum Jump
Forum Jump