Gear / Technical Help > Microphones & Setup

Oddball microphone technique (OMT) - part 4

(1/7) > >>

Continued from part 3..

OMT Part 1 thread-
OMT Part 2 thread-
OMT Part 3 thread-

Will edit this 1st post to add appropriate links as the thread grows.


Ok, back to discussion of the configuration of the L/R near-spaced pair and its role.  I started this post before the discussion of the Improved 3-point PAS at the end of the previous thread, but pushed it off to the start off the new thread, since it sort of recaps a lot of how OMT developed up to this point.  I intend to start a new thread on Improved 3-point PAS to split it off from this more esoteric OMT discussion, or may just append the existing 2-point Improved PAS thread(s).

--- Quote from: Gutbucket on January 11, 2024, 05:35:28 PM ---
--- Quote from: al w. on January 10, 2024, 06:45:39 PM ---[snip..]
Ah, I understand now! I was misinterpreting +/-45° to mean 45° ± Nº (and therefore wondering what range of N to expect) but I realize now you mean +45° on one side and -45º on the other for a total angle of 90° fixed. Thank you.
--- End quote ---

Yes, that's it.  A few more thoughts on that in my next post..
--- End quote ---

Several things going on there.  The imaging relationship across L/C/R mic positions is one.  An improved sense of clarity, proximity, and "upfront impact and bigness" (to my ear, for lack of better description) is another.  The problem is that fully optimizing for one becomes at odds with optimizing for the other.  Over time I've worked on improving the second without overly compromising the first, and I feel this points to a potential convergence between Improved 3-point PAS and OMT I mentioned at the end of the previous thread.

Where this came from (bear with me though a bit of historical progression of OMT)-
I found I gained a lot of flexibility using three mics instead of two, largely because it allows for three new forms of balance control unavailable with just two mic channels: Level of the center verses both sides, slight panning adjustments of the center without affecting the sides, and when needed, the ability to push the L/R energy balance to one side while panning the center the opposite way to compensate.  All things that can be very helpful in the real world of concert taping.  I gained the ability to separate the control of energy distribution and center image position to a limited but useful extent.  Originally this was 3 omnis in mounted in spheres, later a directional mic between two omnis, sometimes 3 directional mics when indoors.  That's 3 ignoring the rear-facing channel or mono SBD in making full use of a 4 channel recorder.

I later increased channel count from 4 to 6 channels by adding Left/Right directional mics between the center mic and the wide omnis, and looked into a number of ways of doing that, trying a few different approaches at my favorite outdoor amphitheater. Günther Thiele's OCT (Optimum Cardioid Triangle) arrangement made a lot of sense to me and I found it worked best for playback over three front speakers, which was my primary focus at the time, but also it worked better for 2ch playback.  Ok good.
The imaging of OCT really does work astoundingly well over three front speakers, which it was originally designed specifically for by minimizing the overlap between the Left and Right mic patterns as much as possible, while retaining just the right of overlap with the Center pattern (not too much, nor too little) so as to hand off smoothly from one segment to the other without conflicting.  It does that by using a cardioid in the center and leveraging the supercardioid pattern in such a way that the null of the fully sideways pointing (+90°) Left supercard is pointed directly at the Right edge of the recording angle, which is where the Right speaker will be during playback, and vice-versa. Pretty cool. In addition to that, sounds arriving from outside of the recording angle window are picked up by the rear-lobe of the opposite mic channel and therefore will be in antiphase, which is sort of like Blumlein.  But unlike Blumlein, as the angle of arrival shifts farther around the sides and on toward the back, the signal into the opposite channel is picked up with much reduced less level due to the the reduced sensitivity of the rear lobe.  That, along with the arrangement capturing phase / time-of-arrival information by being near-spaced, is unlike and arguably perceptually superior to Blumlein.  Its a very clever 3 microphone arrangement.

So my initial OMT array incorporating six channels used an OCT center arrangement with fully sideways facing supercards.  The primary difference between that OMT6 and OCT was the wider spaced omnis and the single rear facing channel.  This worked really well for me outdoors, and much better than most tapers might expect indoors.  I think the directly forward facing center mic anchoring everything, and the other mics being purposefully arranged to work well with it is a big reason why. 

At some point I no longer had a good surround playback system setup, and was listening primarily in stereo.  I still wanted to support excellent surround playback (done right its arguably better used for the replay of live concerts than for anything else), but was really listening most critically in stereo.  First thing I tried was making the center mic more directional, then switched it to a Mid/Side pair. The change to a center coincident pair was all about optimizing 2ch stereo, it wasn't needed for surround. Recording using a 6 channel recorder required sacrificing the single rear-facing channel to do that.  The addition of the coincident center pair worked great for all the reasons discussed at length in previous threads, but I missed the rear channel.  I went back and forth about what was more important.  Really, I wanted both.  Recording 8 channels felt extreme but allowed me to do that.

Also, all this is primarily about general concert taping, and not so much classical music recording done in gorgeous sounding halls where the microphone array can be put in just the right place and shifted slightly until positioned just right. I wanted to better accommodate taper realities, while hopefully also improving things further in already good acoustic situations like nice sounding outdoor amphitheaters where it was already working well.

To do that I wanted the L/R directional pair angled more forward, rather than pointing directly sideways at either side wall (the omnis or side facing subcards cover that just fine).  That change would impart some forward sensitivity bias into the geometry of the array itself, and not just have that forward bias entirely reliant on the level of the center channel in the mix.  To achieve that that I shifted to using a more narrow supercardioid pattern in all three positions and used some additional L/R spacing to try and maintain a somewhat equivalent recording angle to OCT: Supercards in all three positions, a bit more spacing, and L/R angled reasonably forward.  Left/Right at +/-45° is the result of this.  Switching to a supercard center also helped a bit with clarity and a bit more reduction of audience noise in the center, arriving from elsewhere. 

All of that was about increasing forward sensitivity in the L/R directional imaging mics while trying to preserve good imaging.  Upon making the change I found the imaging remained good when listening in stereo.  I began to wonder if I was getting too "belt and suspenders" with phase-correlated imaging being contributed by both the coincident center pair AND the near-spaced triplet.  I found I could use either in the mix, and each provided somewhat different qualities.  The L/C/R tended to be more focused and precise image-placement wise, with cleaner transients, but flatter and less dimensional.  The M/S pair more phase-correlated, yet at the same time more ambient in its stereo width.  I now typically balance both separately in the mix first, then go back and tweak the level of the Side channel while listening to everything in play.  With the near-spaced L/R pair in there providing a lot of the L/R imaging information, I'm using the Side channel as sort of an "ambient width special sauce" type of thing.  It may be more directionally accurate without it, but the right amount of it brings "the sound" that makes me smile and pulls me into the recording.

Because it is directly related to the current topic of conversation, I'm cross posting here what I posted over in Voltronics Solution for time-aligning center mics.. thread this morning, along with a bit of further discussion of whats going on in relation to OMT arrays, and the evolution away from OCT:

Posted earlier in that thread:

--- Quote from: Gutbucket on December 08, 2023, 05:42:21 PM ---The deeper question is what the ideal time-alignment is for the array you are using, and the location you are using it from. Its not necessarily all microphones sharing the same plane (as can be seen when playing with some of the multichannel visual imaging tools, or looking at William's multichannel stereozoom solutions).  But if what is optimal is unknown, then all in one plane is most the logical thing to do.

Optimally managing time alignment and imaging is why the geometry my multichannel arrays tend to be wide with far less front/back dimension.. but some front/back dimension is frequently important and better than having all mics in the same left/right plane.
--- End quote ---

Followed this morning with this..

--- Quote from: Gutbucket on January 15, 2024, 10:54:01 AM ---Back to add some additional detail about center mic alignment, feel free to skip this read if not interested in the technical side of all this-

The classic Decca tree arrangement is sort of a special case as it is essentially placed "inside" the ensemble, which more or less wraps around it.  Sounds arrive at the array from a very wide angle, in essence converging in on it from a half-circle.

In contrast most tapers are dealing with sound arriving from farther out in front of the recording position, in more of a plane-wave like fashion (this is a gross simplification, but bear with it for the purpose of discussion). In that case, there are a couple things going on acoustically.  One is the time-alignment of transient information.  If we really were dealing with a plane-wave, an array of mics placed along a straight line with no forward spacing of the center position, oriented perpendicular to the wavefront arrival would capture it simultaneously in time in all channels.  There would be no time arrival differences across channels. I think this is what most tapers have in mind when they think about time-aligning the center mic position.  Plane waves arriving from directions other than directly in front will produce increasing time of arrival offsets, which will be greatest for a wavefront arriving from fully left or right (+/-90 degrees), as happens with any spaced pair.

Complicating that is the question of whether we really want full time alignment for that front arriving plane wave or not.  Maybe we want to leverage human hearing cues by pinging the center mic position slightly ahead of the L/R positions to strongly  anchor the center in a perceptual sense.

Another aspect is managing the imaging overlap between each pair of microphone channels.  Two spaced microphones will produce a stereo recording angle that is related to the spacing between them (sensitivity pattern and mic angle also play a role, but ignore that for now).   As the spacing between a pair of microphones is made larger, the stereo recording angle grows narrower, but it never actually reaches zero degrees - its edges always splay out wider than the spacing between microphones.  So each pair of microphones produces its own imaging angle, the middle of which which is perpendicular to the line between those two microphones, and the edges of which are angled outward from that center perpendicular line.  If we place all the microphones in a straight line, those outward angled edges will overlap each other.  If we want the edges of each imaging segment to line up nicely with each other instead, so that they hand-off cleanly from one to the next, we need to introduce curvature to the array to angle each segment farther apart until the edges of each segment align.  We can do that in a 3-position mic array by moving the center microphone position forward.

There are complications of course.  Three microphone positions create not just two separate imaging segments but three. We might align the edges of the L-C and C-R imaging segments nicely by pushing the center mic position forward, but the separate L-R imaging segment is going to be considerably narrower than either (because those two microphones are spaced more widely), and will overlap them both.  Leveraging pattern can help manage that. Additionally, it's possible to "steer" the imaging sector of each pair one way or another by carefully leveraging microphone position, time delay, and sensitivity pattern, making it possible to do so asymmetrically where required.  The multichannel Willliams arrays use that technique, some of which require time-delay to get the sectors to line up properly.

It gets complicated!  Not saying what's right or wrong here, just laying out what's going on and the ways in which it can be manipulated.

--- End quote ---

Minimizing any problems of overlap between the L/R imaging segment and the L/C-C/R imaging segments is what OCT achieves by orienting the L/R supercardioids 180° apart, and that management of overlap gets compromised when angling the L and R mic pair more forward.  Spacing them farther apart to compensate for the change of angle helps correct the alignment of the L/C and C/R imaging sectors, yet at the same time increases the level of and aggravates the misalignment of the L/R sector with those imaging sectors.  How much that matters is an open question I don't have a good answer for, but is one reason I've tended to go not narrower than +/-45° with the L/R pair. 

Although I want everything working harmoniously, sufficient clarity and good direct/reverberant ratio are more critically important to me than good imaging.  Because of that, perhaps the optimal solution is going to be improved 3-point PAS, which seeks to maximize clarity and d/r ratio first by pointing the microphones at the source, then secondarily seeks to optimize imaging by determining the spacing between mic positions based on PAS angle.

A question related to all this that still lingers for me..

Given the important role of clarity, perceived proximity and "upfront bigness", which channels of the array are best used for providing those things? 

While I still need to make a more proper comparison between the new stereo shotgun used in place of the center supercardioid M/S pair (by first equalizing each to best effect), upon initial listen I do seem to be getting nicely increased vocal clarity and proximity from the switch to the center shotgun, as posted about at the end of the last thread.

Yet as discussed, earlier I achieved improved clarity, proximity and upfront bigness by angling the near-spaced L/R par more forward.  And that idea might be further extended to pointing the pair directly at the stacks and adjusting its spacing based on that angle (3-point Improved PAS),

So instead of using a shotgun in the center, maybe I should keep the center M/S pair as it was (supercardioid Mid instead of a shotgun) and use short shotguns in the near-spaced L/R positions, leveraging 3-point PAS.

And in that case, maybe the center Mid, being relieved of its up front vocal clarity duties, would be better as a cardioid.. or maybe go the other way and use three shotguns across L/C/R, which could allow for a tighter angle with somewhat less spacing between L and R, since that becomes problematic at narrow PAS angles.

Not sure what the right answer is.  In thinking about this its interesting to frame the question in terms of what the array is supposed to be doing.  I have been thinking of it as "spatially sampling" the acoustic environment, slicing it up into segments which I then use to either feed a surround playback system directly, which is more objective, recreating the illusion of the the sound-field in the room, but is also able to be manipulated via mix-down to 2-ch stereo, a playback format that is more subjective in recreating the illusion of the sound-field for reproduction over just two speakers or a pair of headphones, which may include binaural cues.

I sort of see using a center shotgun as leaning toward the more objective "spatial sampling" approach - the directional sensitivity of the array increases progressively moving around the array to front center, and efforts made to produce excellent 2-ch stereo playback from that is mostly done in post.   Whereas I sort of see using two shotguns in the L/R PAS positions along with a less directional M/S pair in the center as leaning more toward building subjective/perceptual stereo-binaural cues into the array itself.  I'll probably have to get surround playback operating again and so some critical listening to answer this question, by determining how multichannel surround playback is effected by what might only end up benefiting stereo playback..  if it does at all.

Thinking a bit out loud here, but doing so to establish some of the things I'm thinking about as part 4 of the thread gets rolling.


[0] Message Index

[#] Next page

Go to full version