Topic: Why Blumlein sounds more spacious than other coincident or near-so arrays

[Gutbucket]

Blumlein is an odd child in that it seems the only coincident array that really captures space well to my ear.   Of course the imaging is entirely different, but in the spaciousness/ambience department, Blumlein sounds almost more akin to A-B spaced omnis than other coincident or near coincident patterns.   I found some Michael Williams AES papers recently and some new material on Stan Linkwitz's website that have helped me understand the technical reason for why it sounds that way to me. It appears that other than spaced omnis, Blumlein has the highest proportion of stereo reverberation vs mono reverb on playback - meaning the proportion of reverberation that is spread between the playback speakers and not reproduced as a mono source from one speaker or the other. Not to be confused with the direct to reverberant ratio which changes with different patterns and microphone proximity to the source. 

Stan Linkwitz's analysis is here. Of the entire 360degree ambient field, an ORTF arrangement reproduces 48% of that reverberant field as effectively mono, 'hard-panned' to the speaker positions (48% appears to originate from one speaker or the other on playback), where Blumlien reproduces just 30% of the reverberant field that way, the remaining 70% of the reverberation is stereo information spread between the speakers.

Here are his polar diagrams (pickup areas reproduced as mono reverberation are marked in green), for ORTF-

and for Blumlein-


Edit: Here's the significance of the polar plots- All the sound which arrived at the microphone position from within the angles highlighted in green will be reproduced effectively in only one speaker or the other. Reproduction of those regions will be bound to the speaker locations as opposed to emanating from somewhere between them.

Of the limited number of arrays he takes a look at, Linkwitz mentions that the rear lobe of surpercardioid pattern mics allows them to pick up more stereo reverberation than cardioids and extends this observation to figure-8's, noting the ORTF vs Blumlein stereo reverb percentages I quoted in bold above.  Yet he stops short of endorsing the Blumlein array because he is concerned about the folded and reversed imaging of the side and rear pickup quadrants. The supercardioid array he suggests mixes in an additional pair of spaced omnis for reverberant pick-up farther back in the hall. 

Interestingly, Williams suggests intentionally introducing those folded side and reversed rear quadrants for non-Blumlien, multi-channel recordings that are mixed to stereo in his recent 2005 (& 2007) paper below. At the end of the 2005 paper on acoustic cross-talk in multi-microphone arrays, he outlines a possible improvement on the 2 microphone to 2 channel stereo selection by mixing in the 2 additional rear channels in what he calls a twisted-quad mixing scheme that looks quite similar to a the rear-lobes of a Blumlien configuration from the reverberant pickup aspect, but avoids the phase reversed sides.  Doing so spreads the reverberation and reflections arriving at the mic array from each quadrant of the recording space between the two speakers instead of collapsing everything from the far sides into one speaker or the other, but also reverses the back image in a way similar to Blumlien. The 2007 paper goes on to explore mic array design for inter-format compatibility (using one array to simultaneously record for mono, 2 channel stereo, 3, 4, and 5-channel formats) and suggests using twisted-quad mixdown to improve on the 2 channel version.

Also consider that everything above only addresses the horizontal plane (though Williams goes farther in his 1991 paper below to analyze the full 3-D space) and although the mono/stereo reverb ratio of a Blumlein pair shouldn't change with elevation, there is still a sensitivity roll-off and null above and below the mics. An omni pair with perhaps similar mono/stereo reverb proportions would be equally sensitive to reverb, early reflections, etc from those quadrants.
Michael William's page with preprints of his papers and link to Stereo Zoom and Multi-Microphone Array Design tools. 
In particular-
1991 : 91st AES Convention in New York - Preprint 3155
« Early Reflections and Reverberant Field Distribution in Dual Microphone Stereophonic Sound Recording Systems »
2005 : 118th AES Convention in Barcelona: Preprint 6373
«The Whys and Wherefores of Acoustic Cross-talk in Multichannel Microphone Array Design »
2007 : 122nd AES Convention in Vienna - Preprint 7057
«Magic Arrays – Multichannel Microphone Array Design Applied to Microphone Arrays Generating Interformat Compatability»

Stan Likwitz's page on recording for stereo.

[Will_S]

I think a similar argument could explain why I prefer M/S to cards XY, and why I often like hypers.  There too you have directional rear lobes pointed off at angles and should get more of the reverberent field in stereo.

[datbrad]

I think that Blumlein sounds really nice in near field applications, but rarely have I ever heard a Blumlein in far field recording that did not have reverse polarity artifacts from the rear lobes that detract from it.

Since the front and rear lobes are combined in traditional Blumlein recordings, I find them to capture far too much of the room, versus direct signal in PA recording, but that is the only experience I have with them. I do think that today, with a Soundfield, one could make a Blumlein recording in B format and mix it where each lobe is represented by a separate channel, and when played back on a surround system, would replicate the physical space around the mic when the listener is seated in the center of the 4 speakers.

I can think of 2 examples of Blumlein that I experienced, one positive and one negative. In the spring of '94, I patched out of my friend Ronnie's Soundfield at the Phish Beacon run where he ran Blumlein. The bass slap on the back wall, as well as the crowd energy from behind made this a poor choice of patterns, IMO.

In contrast, that summer he ran at the GD RFK shows and in the open air, it was far more pleasing. However, you can still hear more reflections from the rear half of the stadium than an adjacent MK21 recording I have from the same shows.

Application is everything with mic patterns, as PA taping in the traditional sense usually benefits from not accurately capturing the space, at least indoors.

[JiB97]

only time i've run blumlein was June of this year for WSP.  tape came out really nice and clear, but it did lack some low-end, which I had to add in post.  still an enjoyable tape, even if the high end is kinda of grating at times.

[Gutbucket]

I think that Blumlein sounds really nice in near field applications, but rarely have I ever heard a Blumlein in far field recording that did not have reverse polarity artifacts from the rear lobes that detract from it.

Since the front and rear lobes are combined in traditional Blumlein recordings, I find them to capture far too much of the room, versus direct signal in PA recording..

only time i've run blumlein was June of this year for WSP.  tape came out really nice and clear, but it did lack some low-end, which I had to add in post.  still an enjoyable tape, even if the high end is kinda of grating at times.

Good observations and I agree.

Yet I must make a clarification that the observations you both posted above concern different aspects of a Blumlien mic array than what I'm talking about above. They more concern when or when not to choose to use it. Blumlein can be tricky to use and very room dependant. Interestingly, some of the things you guys mention are cases where Blumlien is either similar to spaced omnis (no reduced rear sensitivity so you hear sound bouncing off the back wall) or more like the opposite choice (rolled off bass repose of the fig-8s vs flat omni response down to the lowest octave).  Another problem is that if using it up close it can be difficult to fit all the sound sources in the very narrow recording angle.

I have made some great outdoor Blumlien recordings from farther back were the narrow angle was an advantage, but the 'room' sounded great with nothing but trees surrounding an amphitheater for a really sweet airy ambience - no back wall for the sound to bounce back.  That's pretty much the same situation I would use spaced omnis for. I've run both spaced omnis and Blumlien there simultaneously which is part of what I'm basing my observations on. To my ear, the imaging, bass response and presence are the biggest differences between those recordings.

[anechoic]

excellent post and thanks for the AES papers!!

I've been interested in trying Blumlein for field recording

[illconditioned]

I haven't tried to read the papers yet.

But I have often wondered at the following:  Suppose I have a narrow room, and I put an ORTF or XY pair about halfway back.  I often feel that I'm getting "more wall" than I'm getting sound from the band.  The crazy thing is if I listen from back there (or just use hat mics), it actually sounds better than the cardioids.

My intuition is that the mics are basically pointing at the walls, so if the room is "bright", then I end up hearing a lot of reverberation.  Somehow I'm less bothered by the reverberation when I'm actually in the room.

Anyway, I think Williams is (somehow) trying to explain this effect.

  Richard

[Gutbucket]

Richard, I think you are intuitively on to it.  I've noticed the same thing.  I suspect that the reverberation off the walls is harder to 'hear around' and mentally decipher when it is what is described as mono reverberation above. When the reverberation off the walls is reproduced as stereo information the automatic mental brain processor can better decode the sound stream and hear more clearly 'though' that room sound so you can mentally concentrate on the direct sound.  A bit of the 'cocktail party effect' I guess. 

I've been playing with some surround recording and notice an extension of the same effect.  It becomes much easier to clearly hear whatever you want to concentrate on. In a way it seems that multi-channel recording and playback can 'tolerate' more ambience and reverb while still sounding clear and direct, much like 2 channel stereo can tolerate more ambient reverb than a mono recording could.

excellent post and thanks for the AES papers!!

I've been interested in trying Blumlein for field recording

Thanks should go to Michael Williams who has made all of his papers freely accessible on his website.

[Roger Gustavsson]

For those who understand german, here are some comparsion on different techniques:

http://www.sengpielaudio.com/Stereo-LautlokEines60.pdf

http://www.sengpielaudio.com/Stereo-LautlokEines90.pdf

http://www.sengpielaudio.com/Stereo-LautlokEines120.pdf


Roger

[Gutbucket]

Thanks Roger,
Those are good visual summaries of the angular distortions & totally decipherable even with my extremely limited German.  I like how they are grouped by the angle seen from the microphone perspective which make a lot of sense from a taping perspective since that's how we usually set up - most often picking a configuration to suit the angle imposed by a set distance to the stage, constrained by our position in the room vs the more ideal situation of having the freedom to move wherever we want first.

Richard,
I thought about it some more and it occurred to me that for wider mic angles, even though the Stereo Recording Angle is less, the level difference between channels outside that SRA is greater. I notice in the polar patterns above that the mono reverb areas are regions where the level difference between channels is greater than a certain threshold amount.  Linkwitz notes on his site that using a wider spacing and corresponding narrower angle between the mics for the same Stereo Recording Angle increases the stereo reverb proportion, which is in-line with that since the level difference between mics is less at any angle outside the SRA, by an increase in the time/phase difference.

That makes me wonder what the 'worst' configuration would be that maximizes the mono reverb proportion.  I suppose it would be cardioids angled 180 degrees apart, so that the null of one corresponds with the primary axis of the other and everything outside the narrow SRA to the front and back would exceed the level difference threshold.  That insight also helps me to understand why widely spaced omnis with a big hole in the middle and the playback image of sources pulled toward one speaker or the other still sound stereo reverberant - the phase/timing difference between channels draws the apparent source positions toward the speakers (which at first seemed similar to the mono-reverb situation to me) but the level difference between channels is not great enough to exceed that threshold.  But that's all just my suspicion at this point.  I need to go back and re-read to look for that threshold amount.

[Chilly Brioschi]

No discussion of Blumlein can be complete without mentioning the Kentucky Colonel's Appalachian Swing!




1st, yes that is David "Dawg" Grisman in a bolo necktie, white collared shirt and spiffy red jacket!

2nd, there was use of Blumlein and multiple players on an array  (a bit experimental) which literally spins in the mix, like a ride at Disneyland

3rd, I'm just kidding about Grisman, that's actually Roland White on the album....     

4th, I'm not kidding about the Blumlein and dizzying sense of space and placement

[run_run_run]

No discussion of Blumlein can be complete without mentioning the Kentucky Colonel's Appalachian Swing!




1st, yes that is David "Dawg" Grisman in a bolo necktie, white collared shirt and spiffy red jacket!

2nd, there was use of Blumlein and multiple players on an array  (a bit experimental) which literally spins in the mix, like a ride at Disneyland

3rd, I'm just kidding about Grisman, that's actually Roland White on the album....     

4th, I'm not kidding about the Blumlein and dizzying sense of space and placement

looks epic

[Chilly Brioschi]

looks epic

It's Bluegrass standards, played totally kickass, and no vocals.
No true bluegrass fan should be without it.
The Colonel's are the article, but often overlooked.

[Gutbucket]

^^^
I'm imagining the band dong the classic bluegrass, 'band around a single mic dance', lifting instruments up for solos and accents, sliding to the side to make room for the featured instrument, crowing up for harmonies - an amazing coordination of movement.  I always love seeing that both for the purity of the sound, the self-mixing of instrument dynamics and voices by a talented band as well as the visual entertainment - its a wonder that they can get in there and out of the way in time.  I saw a guest mando player get knocked on his ass once, hip-checked by the banjo player when he didn't get out of the way in time - quite a crowd pleaser, but both the banjo player and mando man were red faced.

If that's the case on that album, the dizzying flying instruments could have the same effect as a trip down the blue ridge parkway for a motion sick-prone kid in the back seat.  Quick pull over!

Thanks for the tip Bob, I'll have to search that one out.

[hugh]

Interesting stuff.  Plenty of good reading in those links!  And that bluegrass record looks worth tracking down :-)

Here's another paper about reverberation and image solidity.  It basically recommends 90 degree separation for figure-8 and 120 degrees for cardioid, based on the most equal distribution of reverberation across the soundstage.
http://www.audiosignal.co.uk/Resources/Stabilising_stereo_images_USL.pdf