Quasi-binaural setup

Getting closer to the final aim of my investigations, this post will present some investigations on combining the microphone horn addons with a dummy head to form a quasi binaural recording setup.

Overall objectives for the setup are:

• Good and clear pickup of sounds from the surroundings
• Good stereo imaging in headphones, but also usable with speakers
• Adaptable for various recording situations
• Portable (on a bike)

We are in the field of binaural recording techniques. A lot can be read, and a good place to start is wikipedia:

• Binaural recording
• Dummy head recording
• Head related transfer function
• Sound localization

The overall objectives can be broken down into a set of stereo imaging and frequency response issues:

Stereo imaging

• Delay effect between each channel corresponding to effect with human ears
• Volume panning between left and right
• Directional coloration to emulate back, front, up and down sensation

The frequency response may be optimized depending of recording situation being

• Long distance recordings – will need HF lift
• High frequency, ultrasonic recordings – will need linear pickup from +15kHz and up
• Music – will need quite linear lower frequencies, maybe with a bit of warmth
• Near field – will need linear pickup
• Speech – will need emphasized presence

Concept

So my approach is to use a combination of a horn addon and a dummy head.

and the result is…..Rock star Dummy

Rock star dummy

The dummy head will provide the natural time delay between the ears and also a great deal of left-right volume separation.

The horn addons will do additional focusing and add a little of directional coloration. The horns will also emphasize a given frequency band according to the recording situation.

Tests

Using the directional testing method from the earlier post Horn directionality it is possible to get at good overview of the frequency response and directionality.

Dummy head with one clean microphone at the rotation axis

Side view of the dummy head

Top view of the dummy head with one microphone to the left

The dummy is constructed using a yarn bobbin of the same size of a human head, covered with a textile top layer. Different types of top layers can be applied to get a desired acoustic effect. On the pictures above an additional collar of fibers is added to further reduce reflections.

The microphone is placed outside the center o the head when seen from the side.

The images above show a typical test setup which can rotate keeping the microphone tip at the rotation axis. Noise was sent out by a speaker 78 cm away, and the setup was rotated 360 degrees in 60seconds. Again, the post processing was done using CoolEdit Pro to generate a spectral view, which following was converted into polar coordinates in GIMP.

Results

The charts below show the results for different dummy setups. The varying parameters are the distance between the microphone and head, different horn addons, and different materials on the dummy head.

1. Clean microphones, blank textile 2.2 cm from surface

Here the dummy head has blank textile top layer. Clean microphone placed at the side with a distance of 2.2 cm to the side walls.

Clean mikes, blank textile 2.2 cm

This setup clearly suffers from strong reflections from the side walls giving a dominating comb filter effect for sounds coming from the free side. The minima corresponds well to d=(2n-1)*lambda/4, ie lambda=4d/(2n-1)=4d/1, 4d/3, 4d/5 … etc.

This setup has very a small difference between front and back.

2. Clean microphones, blank textile 2 mm from surface

Same setup as above but with a distance of 2 mm to the side wall. Again strong reflections occur from the side wall, but the first minimum of the comb filter is now around 14 – 19 kHz, thus much less of a problem. This setup resembles a boundary type setup.

Clean mike, blank textile 2 mm

3. Clean microphone, wool+synthetic fur top layer, 3 cm distance

Dummy head with woolen top layer and fur sides. Clean microphone placed at a distance of 3 cm.

Clean mike, woolen hair 3 cm

This top layer has a good acoustic absorption and effectively eliminates reflections compared to the blank textile. Tiny ripples can however still be seen for sound coming 90 degrees from the free side.

The spectral image is very homogeneous in all free directions, especially for frequencies up to 22 kHz.

4. Clean microphones, wool+hair close

Same as above but with a minimum distance, almost touching the side wall.

Clean mike woolen hair close

Putting the microphone closer to the side walls with this setup only slightly changes the result. The shadow side is clearly increased.

5. Toothpick horn, blank textile close

Here the dummy head only has the blank textile top layer and the microphone is fitted with a toothpick horn which is almost touching the sidewall. The toothpick horn shape is flat sloping outwards, pointy side. Effectively the microphone tip is about 19 mm from the side wall.

Toothpick horn, blank textile, close

The reflection pattern from the side wall dominates the frequency response with a comb filter effect. Due to the horn an asymmetric pattern is seen, with a larger difference between front and back.

6. Toothpick horn, wool+fur, close

Dummy head with woolen and fur sides. Again microphone with toothpick horn, almost touching the sidewall.

Toothpick horn, woolen hair, close

This setup has a good reception for sounds coming from the front (0 – 45 degrees). At 90 degrees angle a minimum is found at 17 kHz. At even higher angles frequencies above 2 kHz are attenuated gradually. A bit of presence boost is found for sounds coming from the front.

toothpick horn geometry

This setup has good reception from font angles and gives a fine coloration as the sound is turned around. I believe this should give a good directional sensation.

Conclusion

The surface material of the dummy head is very important and can potentially have a high influence on the received directional response. If the material opposite the microphones is reflecting the sound waves, negative and positive interference occurs (comb filter). When the sound comes right from the side, the first negative interference occurs when the distance is equal to 1/4 of the wave length, the following at 3/4,  5/4,  7/4 etc. To avoid this, a sound absorbing material must be used.

I think the the final setup no. 6 shows the most promising behavior for nature recordings due to the good front side presence and the gradual coloration when rotating towards the back.

Setup no. 2 has a more front-back symmetric pattern with a high midtone boost for sound coming at 90 degrees.

Setup no. 3-4 with clean microphones may be useful if a more flat and omni directional frequency response is needed.