Posted by: recordingsofnature | December 25, 2015

Simplified microphone design to mimic binaural forward elevation cues

In this post, I will try to come up with and test a simple microphone design which can produce certain directional cues to enhance the spatial feel of stereo audio recordings, particularly the forward up and down sensation.

Stereo recordings are normally based on volume level difference and time difference between the microphones. Typical used techniques include ORTF, M/S setup eventual combined  with a baffle between the microphones (such as a Jecklin disk).

Information about elevation is however largely missing with these stereo methods. In principle, it is not directly possible to tell if a sound comes from above, down, front or back. Binaural microphone techniques (using anatomical ears and dummy head etc.) are able to provide full directional sensation by reproducing the delicate spectral coloring (HRTF) from the human ear and head/body, which tells the brain about the direction of the sound. But this method has a few drawbacks in terms of personal differences and dominating resonances which need post equalization.

So, here I will try to develop a simpler microphone shape which can mimic the up/down directional sensation of the binaural method to improve the forward up and down sensation of my nature recordings.

Characteristics of forward, up and down directional sensation

The characteristics of the forward elevation cues can be studied with the Head related transfer functions (HRTFs). An example is given below which shows frequency responses for different directions in the horisontal and vertical plane (reworked from Dmitry N. Zotkin et al. . Acoust. Soc. Am. 120 4, October 2006)

hrtfs overview

The forward HRTF spectra for the side and for forward elevation are summarized below.

theoretica hrtfand below are the comparable results from estimating my own HRTF.

kr hrtf

Please note, that the charts for forward elevation mostly are calibrated to the average (diffuse field) response, so the relative differences becomes more visible.

From the charts, I think it fits relatively well to say, that the forward elevation cues are controlled by 8-9 kHz notch, caused by some quarterwave reflection in the ear of forward coming sounds. This notch goes lower (7kHz) at -45 degrees elevation, and moves higher (into nearly inaudiable frequencies, 12kHz) for upwards directions. Directly above the overall response is fairly flat. The notch also gradually disappears when going to the side, at 90 degrees.

In general, most changes in the HRTF are happening when the sound hits the ear with a direct point of view. In the shadow zone the changes are more subtle and an acceptable approximation could be to aim at a flat response in this area.

Test procedure

In the tests, I will gradually build up the microphone starting with a flat structure and the add features, including a lobe 14-17mm behind the capsule to create the forward 8 kHz notch.

The test setup, as seen below, uses the tree-ears microphones mounted on each side of a dummy head stand, in living room conditions.

IMG_6069  IMG_6146





The reference noise is produced by rubbing a thin plastic bag. For each microphone setup I move the noise source around and up/down in the same rote. First, going around the setup in 45 degrees increments, then up and down in a 45 degrees vertical circle in front and finally forward down, up and to down to the side. The route can be seen below:

test schemeRubbing the plastic bag produces a nice smooth noise spectrum. All the recordings are following analysed in Cooledit and with a script in Scilab, where the noise spectra are compared with the reference noise recording  (1. Clean QTC 40).


The aim with the following sound samples, is to let the listener make a subjective evaluation of the sound, especially on the perceived direction. Headphones are needed to hear the effects.

1. Clean QTC40 AB setup

clean qtchrtf_rawdata

This is a clean reference recording using a simple AB15 setup with quality qtc40 omni microphones. This setup provides the reference noise spectrum for which all the other measurements are calibrated against. Note how this setup has quite limited directional feel.

2. Tree ears platform with flat ear

clean 6.23 esponse

Here the Primo 172 microphone capsules are just mounted at an almost flat surface, with no features to affect the sound. This should give the most clean sound with the tree ears platform. The curved course of the frequency charts is most likely a results from the “tree trunk”. Note how this produces very limited sense of up and down, e.g. to my ears sound from directly forward appears coming from an upwards direction.

3. Tree ears with small tube

small tube response

Small clay tube 3.5mm thick is placed just behind the capsule to give a bit >9kHz gain. Still very limited elevation sense, same as the previous. Very little up/down sensation.

4. Tree ears with small tube and small back lobe

no vax response

Configuration with the small tube and a larger curved back lobe, starting appx 14-17mm behind center of capsule and 5mm high. The lobe is to create a reflection of forward sounds, which also creates a notch around 7.5 kHz. This can also be seen on the charts. I think forward sounds now sound more in level, but still a bit up. Start to sense upwards directions better, but up and down completely confused.

5. Tree ears with small tube and larger back lobe

concave response

The back lobe has now been extended with clay to a height of 8mm and the notch is now stronger. Still the response from the side and in the shadow region is relatively flat. Sounds from forward sound more in level now. Response for up and down is however still somewhat confused and nearly the same.

6. Tree ears with anatomic ear (ne34)

mulitchannel treeears

Here is a comparison with a setup with the microphone capsules inside anatomic ear models. The sound is not equalised, so not a very pleasant sound. The up/down response now looks closer to the reference HRTF. I think it is now possible to distinguish up and down in the sound sample.

7. Binaural setup


Equalized version

st binaural response

This is a comparison with my binaural setup. Frequencies now looks closer to target response. Maybe the overall frequency color is not the best, why I have tried to compensate in the equalized version. To my ears, there is a quite clear directional sense for forward, backwards, and elevation.


Im aware this is a very subjective test. I’m curious to know if other listeners have the same perception at all, as I have.

To my ears, adding the back lobe, which creates the 8kHz notch have a clear effect of moving the forward sounds downwards, and makes it possible to distinguish forward from up . Without the notch, forward sounds sound like coming from 45 degrees up.

So, I think the setup with the larger back lobe (5) gives the best performance, with a good sense of forward elevation (though not downwards!). Other directions have relatively flat response, which gives an overall smooth sound.

The down direction is not really successful with any of the simple setups.  It will need extension of the back lobe, but the risk is, that the ear design will end up as a complete ear. If needed this will be next step. It could be quite fast to do experiments using clay.

It is clear to hear that the noise sound varies quite a bit with different setups. The anatomical ears have a very colored and resonant sound compared to the more flat and simple designs. On the other hand, I also believe the brain is very good at compensating when listening over a few minutes. Then is it the relative differences that provides the directional sense. It may all be a matter of finding the right compromise.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: