The World's Finest Blumlein Array

[Len Moskowitz (Core Sound)]

One of the most popular microphone techniques for recording vibrant. precise and accurate stereo images is the Blumlein array.

It's two dipole (figure-8) microphones crossed at 90-degrees and centered with one at +45-degrees, and the other at -45-degrees.

We say that a single Core Sound's OctoMic (a second-order ambisonic microphone) is the world's finest Blumlein array.

What makes it so? The OctoMic's virtual dipole microphones' polar directivity patterns make it so.

Here they are:

 
 

Please compare them to any world class dipole microphone from any manufacturer in the world. You'll see that OctoMic preserves the dipole pattern up to higher frequencies, down to lower frequencies, and with smaller deviations from a perfect dipole.

(Note that these graphs uses a linear scale to show off how consistent the OctoMic response is. Comparing to log scales, 0.5 on these graphs is the same as 6 dB down on the log graphs. 0.25 is 12 dB down.)

Rotate the two patterns 45-degrees in post-production and you have the world's finest Blumlein array.

Hear an OctoMic Blumlein array recording of Jefferson Airplane's song "Somebody to Love" played by Back To The Garden 1969: https://u.pcloud.link/publink/show?code=XZ0coNkZPaeBJB24WkYeXnjPuyTpMXb9PA6V

Hear an OctoMic Blumlein array recording of Stile Antico at NYC's Music Before 1800, recorded by Rob Anderson, here: https://u.pcloud.link/publink/show?code=XZP7yakZXjfMVKMBIAysJHbFhlmWc8NiYQh7

Hear an OctoMic Blumlein array recording of the Allman Brother's "Midnight Rider" played by The Friends of the Brothers here: https://u.pcloud.link/publink/show?code=XZ2yyakZrEzyjXFa2DkdUdsXUgWyPzUyJId7

Learn how to use OctoMic as a Blumlein array here: https://www.core-sound.com/OctoMic/4.php#Blumlein

Learn more about OctoMic here: www.core-sound.com/OctoMic/1.php

See more OctoMic polar directivity patterns here: www.core-sound.com/OctoMic/2.php

[EmRR]

Very interesting, particularly that 10kHz gets more sensitive towards 90º where most get less sensitive with increasing frequency.  I suppose the 10kHz variation comes down to the head shape and spacing (as in every mic), with a result differing from traditional small figure 8's.  I'm curious how it (or any other) looks even lower, say 50Hz, though pretty much no company show data down in that range.  Not to detract in any way, this looks great.  Hope to try one someday.   

[Gutbucket]

Noticed the same trend up top.  As I'm sure you are aware, rather than being native the bidirectional pattern (as well as all others) is derived, so there is significantly different stuff going on with respect to the acoustic capsule geometry relationships in comparison to standard single and dual diaphragm fig-8's.

Len, as a TetraMic user (for those following along, that's CoreSound's 1st order, 4 channel ambisonic microphone which proceeded the 2nd order 8ch OctoMic), I'm curious how much the TetraMic bi-directional pattern differs. The TetraMic is capable of a well behaved Blumlein pattern itself. Are the two similar with regards to the behavior of all 1st order patterns? Or is the 2nd order capability able to be used to demonstrably improve upon the 1st order pattern behavior of the TetraMic? 

What most interests me about the OctoMic is potential for deriving directional pickup patterns unavailable to 1st order microphones.  Specifically cardioid-like patterns with a narrow front lobe similar to that of a 1st order figure-8, yet with minimal back lobe sensitivity (and acceptably minimal side lobe behavior).  Can you tell us more about those aspects?

With the TetraMic in typical taper use, I generally found myself gravitating toward dialing in crossed supercardioid to hypercardioid patterns with a somewhat wider included angle rather than Blumlein.  That's probably not surprising to folks here intimately familiar with the use of coincident pair techniques in taper situations. It is in that pattern range where I expect the advantage of a higher order system to offer some interesting possibilities and the potential of significant improvement.


BTW, that Stile Antico recording is very nice.

[Len Moskowitz (Core Sound)]

You can see more OctoMic polar patterns here:

www.core-sound.com/OctoMic/2.php

As you can see, while a first-order cardioid is down 6 dB at 90 degrees, a second-order cardioid is down 12 dB. Look at the depth of the nulls at 180 degrees.

And OctoMic's patterns' stability over frequency is phenomenal.

You can get much tighter patterns, if that's what you want. Second-order mics will give you patterns not available from mono mics. You can see some of them on the page noted above.

While TetraMic's polar patterns are probably the best of the first-order microphones - because we calibrate and correct each one of them - they don't compare to OctoMic's.

[Gutbucket]

While TetraMic's polar patterns are probably the best of the first-order microphones - because we calibrate and correct each one of them - they don't compare to OctoMic's.

More specifically, I'm especially curious how much difference there is between the two if restricting the comparison to 1st order patterns in the super/hyperish-cardioid range.

[Gutbucket]

With regards to 2nd order patterns, are there any recordings available using them?  Coincident stereo configurations in particular where the behavior of 2nd order pattern interactions are less than intuitive?

Will be interesting to listen for how 2nd order hypercardioid patterns interact in a coincident stereo pair with regards to the negative-polarity side-lobes and the positive-polarity rear lobe. 


Looking at the plot above it appears a pair of 2nd order hypercardioids with something like a 75 degree inclusive angle may achieve a panning law similar to a crossed pair of fig-8s with a narrow 45 degree inclusive angle, and overlapping out of polarity information outside of it.  A PA/stage width "orchestra angle" of 75 degrees or less is common for audience recording positions, but I'm not sure at all how that would translate or how useful it might be.

This gets me thinking about the possibility of tweaking intermediate pattern shapes between 1st and 2nd order (selecting an optimal balance by ear), is such a thing possible with OctoMic?

[Len Moskowitz (Core Sound)]

More specifically, I'm especially curious how much difference there is between the two if restricting the comparison to 1st order patterns in the super/hyperish-cardioid range.

There's a lot of difference. The OctoMic's higher order spatial sampling results in the consistency of the patterns.

You can play with the various polar patterns yourself, at both first-and second-orders. Download some of the B-format files from the OctoMic recording page, and decode them yourself with the SPARTA beamformer decoder.

SPARTA plugins: research.spa.aalto.fi/projects/sparta_vsts/plugins.html
http://research.spa.aalto.fi/projects/sparta_vsts/img/Beamformer_GUI.png

OctoMic recordings: www.core-sound.com/OctoMic/13.php

The SPARTA beamformer has just a few selections, but they're quite useful. There are more interesting and flexible beamformers being developed.

[Len Moskowitz (Core Sound)]

With regards to 2nd order patterns, are there any recordings available using them?  Coincident stereo configurations in particular where the behavior of 2nd order pattern interactions are less than intuitive?

Will be interesting to listen for how 2nd order hypercardioid patterns interact in a coincident stereo pair with regards to the negative-polarity side-lobes and the positive-polarity rear lobe. 

Here's an interesting comparison of four different stereo virtual microphone decodes of a recording made with one OctoMic ambisonic microphone. They are all decodes of a recording done with a single OctoMic.

(OctoMic is our second-order ambisonic microphone. Each one is individually calibrated, and can be decoded to multiple virtual microphones having very precise and consistent polar patterns. It can also be decoded to full 360 surround and headtracked binaural )

The recording is of Angelica Women's Chamber Choir (angelicavoices.org) singing "Hic vir despiciens mundum" at Church of  St. John Nepomucene in NYC (May 19, 2019).

The chorus was arranged in a semi-circle. The church has a reverb period of longer than six seconds, and is situated on a very busy (and noisy) NYC street.

The decodes are:

1. Blumlein (two figure-8 microphones pointed at +45 and -45 degrees.

www.core-sound.com/temp/Hic-vir-despiciens-mundum-Blumlein.wav

2. Two second-order hypercardioids pointed at +45 and -45 degrees

www.core-sound.com/temp/Hic-vir-despiciens-mundum-Two-Hypercardioids.wav

3. Three second-order hypercardioids pointed at +45, 0 and -45 degrees, with the center microphone mixed at -3 dB

www.core-sound.com/temp/Hic-vir-despiciens-mundum-Three-Hypercards-to-Stereo-center-3.wav

4. Three second-order hypercardioids pointed at +45, 0 and -45 degrees with the center microphone mixed at +6 dB

www.core-sound.com/temp/Hic-vir-despiciens-mundum-Three-Hypercards-to-Stereo-CENTER+6.wav

[Len Moskowitz (Core Sound)]

Here's a recording made with an OctoMic and decoded to Blumlein.

The performers are Stile Antico at Music Before 1800 in NYC.

https://u.pcloud.link/publink/show?code=XZP7yakZXjfMVKMBIAysJHbFhlmWc8NiYQh7

Listen to the incredibly sharp imaging of the choir in their semi-circle.

Thanks to Rob Anderson for the recording, and to all involved for permission to post this recording. Photo courtesy of The NY Times.

[Gutbucket]

Thanks for the links to these samples.

[voltronic]

Very nice samples, Len.  I'm a choir director, and this music could not be more in my wheelhouse.  I also love Stile Antico - own several of their albums.

The most convincing for me was the straight-up Blumlein of the Anglica Women's Choir, as it captured the acoustic the best while still giving very good balance to the choir.  That's really impressive - the best Blumlein recordings I have heard usually use single-diaphragm fig8s like MKH 30 or MK 8, and there are some that it's verboten to do it any other way (I'm not one of them).

Second place goes to the two second-order hypers.

For a recording such as this, I always like to try spaced omnis 40-60 cm, but that's the one thing a single ambisonic mic cannot give me, I suppose! 


For the Stile Antico selection: was this room extremely dry and/or with a low ceiling?  The imaging is indeed pinpoint, but sounds like X/Y cardioids in that I am not hearing any room reflections (if there were any to be had).

Thanks for sharing this.

[Len Moskowitz (Core Sound)]

Stile Antico was recorded at Corpus Christi Church in NYC.

You can see photos of the sanctuary here:

https://www.corpus-christi-nyc.org

You're hearing the room, but it doesn't add much. It's a medium-sized colonial-style wood church, not a large stone church like St. John Nepomucene. The ground floor pews were full. The balconies are on three sides - they were empty. The room is pretty much a cube with the chancel on the north side of the room.

I probably have a recording of a loud hand clap in the empty hall somewhere. Of course, the acoustics were much dampened by the audience.

[voltronic]

Thanks again.  I had a feeling it was just a dry acoustic.


Regarding the two omni samples:

These are the least successful to my ears.  I admit I don't really have the first idea of what goes on with Harpex, but on the Angelica example, the image width has collapsed and become much more distant, as though a directional stereo array were rotated around so the rear of the mics were facing the choir.  In the Stile Antico example, the L/R placement of the choir sections has been flipped and there is also a persistent buzz not present before, so I wonder if something went amiss in the decoding of that track.  I hope you don't mind the criticism; I mean it to be constructive, but I don't think those two examples are indicative of the performance of your mic.


I continue to be very impressed with the other examples, especially the Blumlein decodes of both choirs.

[Gutbucket]

Looking forward to listening to all of these.

Volt- I've been long curious about Harpex's virtual patterns and stereo output derivations but have never gotten around to demoing it using any of my own recordings.  Especially curious about it's synthesized near-spaced and A-B configurations, which I presume use the plane wave decomposition function Len mentions, yet I remain skeptical about how well real-world phase/spacing interactions can be simulated using sample data from only a single point in space.

[Len Moskowitz (Core Sound)]

I too am not very happy with Harpex's synthesized spaced mic decodes. It does a good job with first-order recordings that are not complex, though if I need a parametric decoder, I use COMPASS.

I've removed those two recording links.