Thanks for posting the recordings Heathen. Its informative to be able to compare with the two other recordings. Listening around the general frequency response differences, which is almost always the first thing I notice and recognizing that influences perception of other aspects, I hear more 3 dimensionality in your recording which is one of the things I've come to expect from an OMT arrangement. Interestingly, I notice more audience pickup in the Ricky's PAS stereo pair. That's somewhat surprising as I imagine the PAS angle from the recording position you both shared was likely significantly less than the 110° used between your AT's.
This brings to mind something I find useful in considering how an arrangement of microphones works in a collective sense. Not with regards to stereo image pattern overlap, but in terms of overall sensitivity and how that sensitivity is distributed. I've posted previously about the
overall sensitivity pattern of an array. By which I mean the collective combination of all microphones in the array in sort of a monophonic sense, at least in that they overlap each other and produce a collective sensitivity pattern "shape". We modify this collective sensitivity pattern before recording by choosing the patterns, angles and positions of the microphones which form the array. We further modify the shape of that pattern afterward by choosing which channels to include and adjusting their levels relative to each other. One thing adjusting the level of the center microphone channel against the Left and Right microphone channels does is modify the shape of the collective sensitivity pattern of the array.
The entire picture is considerably more complex, but this way of thinking is a useful simplification for illuminating the most fundamental aspect of an array's overall sensitivity to the 3d sound field in which it is immersed at the recording location.
On the Sengpiel Audio stereo microphone array visualizer [
http://www.sengpielaudio.com/HejiaE.htm], the collective sensitivity is represented by a single grey pattern line which surrounds the two smaller, individual microphone pattern lines in red. When initially loaded, the virtualzer defaults to an EBS microphone setup, which is a pair of cardioids spaced 25cm and angled 90° apart from each other. The polars of the individual microphones in red are cardioid shaped. The collective sensitivity pattern of the two in combination is shown in grey and also cardioid shaped. It's actually something of a directly forward-facing subcardioid. If you change the angle between mics from 90° to 0° the collective pattern tightens up to a true cardioid.. if you change the angle to 180° it widens into more of a circular shape, indicating a more omnidirectional collective sensitivity for the combined array of two cardioids facing away from each other.
I'm thinking of recording in a room as a "sensing problem", first simplified to this kind of overall collective sensitivity shape (sort of the monophonic case if you will), then further differentiated with a careful ear towards the interactions between whatever divisions we choose to further impose on it.
Like a single omni, a Blumlien 90° crossed pair of fig-8's, or an ambisonic microphone (each of which have equal sensitivity across all horizontal directions) most multichannel OMT arrangements also have sensitivity across the entire horizontal plane. That's because they commonly include omnis and/or include directional microphones that are generally pointed so as to cover all cardinal directions, at least to the degree that their patterns overlap and provide sensitivity to direct sound arrival across all horizontal directions. We gain some control over the shape of that collective horizontal sensitivity by way of the physical arrangement of the microphones. We further modify that collective sensitivity pattern by choosing which channels get used, their routing (which get assigned to the Left output channel, which to the Right, which to both), level adjustment, EQ or other filtering, Mid/Side ratio adjustments, and perhaps subtle panning adjustments as necessary. The key is that we gain deferential control over sound arriving from different portions of the horizontal plane to some degree, both prior to and after making the recording.
The stereo and 3d cues are generated by differences between the microphone channels. In addition to those cues picked up from the environment itself, they include differences in phase, time-of-arrival, and response due to the particular microphones used and how they are positioned and oriented with respect to each other. In other words, they are both "of the environment" and "of the array" in terms of the way in which the array is energized by the sounds in environment (and how we further modify the interaction of channels in the mix). Such "of the array" interactions are inevitable, so we should make sure the interactions between the microphones themselves make for positive, or at least neutral additional cues which don't end up doing more harm than good. That gets tricky.
Binaural human hearing is similar in that it is collectively omnidirectional in the overall sense, yet differentiated in such a way that we are able to determine direction and other auditory attributes. Similarly, the raw data contains cues that are both "of the environment" and "of the interactions of our sensing array". The cues are complex, not simply level and time-of-arrival based. Our brains, formed by evolution and trained by life experience filter and interpret the data in such a way as to consciously ignore the base-level interactions particular to the sensing array of funny shaped ears placed on opposite sides of an actively articulated head - the end result being conscious perception of directional cues and other high level auditory aspects, without noting level, spectral, and timing differences directly. It either works right, or it doesn't. If it doesn't something may seem "off" and to some extent we can train ourselves to be more consciously aware of such things, or it might just sound bad or lacking or flat or involving or whatever.
We do well to design our artificial hearing array mic setups in such a way as to play nicely with the expectations of our brains. And by play nicely I mean trick ourselves into believing the auditory illusion is convincing. Beyond the ability to stack the deck by devising clever microphone arrangements and manipulating the resulting channels in the ways we do currently, there are a clear paths forward toward creating more sophisticated auditory illusions using the recordings we are making today. I can already do that to some extent by manipulating for playback over more than two speakers, using relatively simple forms of multichannel surround playback. But I also know of and foresee various ways of making more convincing illusions over 2 speakers or headphones by more complex manipulations of the differentiated data we collect, making for a better fit to the complexities of the human hearing sensory array and producing an even more convincing high-level conscious experience.
I might be reaching a bit, but I think the careful adjustment of the center mic channel level verses the other two in heathen's recording represents an important step in that direction. Consider how adding a second microphone channel to mono is able to change our perception so dramatically by producing a far more robust stereo illusion of the original event. The big change is not the addition of another channel" but what our brain does with the additional information it is able to provide.