by tomlinson holman
“Multichannel Natural Music Recording based on Psychoacoustic Principles” by Dr. Günther Theile of the IRT in Germany is a 37-page monograph with lots of both new and tutorial information about recording multichannel sound. It is available as a PDF off the Internet, from www.irt.de/IRT/indexorgani.htm/. Theile is certainly one of the preeminent researchers in this field. His employer, the IRT of Munich (Institut für Rundfunktechnik — Institute for Broadcasting Technique) is one of the only organizations in the world supporting multichannel research, and so both parties should be commended for this work. While we always try to make Relevant Research really relevant to surround professionals, sometimes we cover topics of interest to a narrow range of our readers. This isn’t one of those times. Virtually anyone who reads this magazine seriously should download this work and read it if you want to know what’s going on with various methods of real multichannel miking (as opposed to multitracking and panning everything into position — not that there’s anything wrong with that).
The work first seeks to define naturalness, and its optimum qualities. “The desired natural stereophonic image should therefore meet two requirements: it should be satisfying aesthetically and at the same time it should match the tonal and spatial properties of the original sound.” Of course, this isn’t everyone’s goal, because the original sound may be completely artificial, or never existing except in a producer’s mind. Still, there are valuable lessons here for all practitioners to have in reproducing a real event as a basis for discussion.
Section 2 is on LCR stereophonic imaging. This walks through various microphone setups, including the five-across-the-front style I’ve used sometimes, and the problem with that system, due to interchannel crosstalk among the omni or cardioid microphones. Theile also describes a “triple phantom source” problem arising with just three mics.
He then describes a number of different widely used configurations from near-coincident to widely spaced omnis in terms of their imaging capability. The analysis includes easy-to-read curves of the various systems. For instance, the INA method proposed by French researcher Williams produces imaging that makes the curve in fig. 1.
This shows a rapid change in direction about the centerline, not the desired smooth change.
The well-known Decca-Tree consisting of spaced omnis makes the curve shown in fig. 2.
Near-coincident techniques (fig. 3) are a little better. In this case they consist of three microphones spaced at 17.5 cm x 2 (6.88 in x 2), with the outer two being supercardioids and the inner one a cardioid, and the outer ones rotated outwards by 30 degrees. This helps to prevent crosstalk among the channels.
Perhaps the most interesting development goes beyond study of these conventional systems. Theile’s Optimized Cardioid Triangle offers potentially the best imaging diagram (fig. 4).
In fact, there’s an interactive application that runs on the Web to show off the newly proposed system variations, producing curves like those shown here at www.hauptmikrofon. de/ima-folder-eng/ image.html. Try it: it’s fun. You can move the microphones around and change their types, and the output curve shows what imaging will be like for this combination. The fig. 5 shows one simulation I ran that produced pretty good results (where the Y (vertical) axis is the excursion in degrees and the X (horizontal) axis is the angle of incidence in degrees).
Of course, in this system the main microphones are very far off axis from front sources, and must be equalized so that, say, their 30-degree incidence is flat, which will require some high-frequency shelving boost equalization.
Theile goes on to describe how to extend the bass response, since the pressure-gradient microphones used in this system roll off at extreme bass frequencies. This means adding in an omni with a crossover for the lowest frequencies.
The next section describes depth imaging in detail and the related effects of spatial distribution of reflection patterns. Section 3.1 is devoted to lateral stereophonic sounds that basically don’t work in the standard 5.1-channel setup due to the huge head-related transfer function stuff going on. “It is not evident that stereophonic lateral areas can be suitably applied for this purpose, since the loudspeaker arrangement…is not intended for lateral placement of stable sources and for presentation of completely surrounding and localizable auditory events. It is well known in theory…and from a number of experiments…that lateral phantom sound sources are extremely unstable and sensitive regarding signal spectrum and listener’s location.” Theile’s answer is the same one that we have in the 10.2-channel system: add channels intermediate between L and LS and between R and RS and you find that side images do work as we reported on recording Shakespeare “in the round.”
Then surround microphone setups are discussed. A square arrangement of microphones is proposed for OCT to produce good spatial imaging and envelopment. Added to the three front mics with adjustable time delay (a main reason to sell your boss on a new digital console), the product is said to be very good in imaging and envelopment.
All in all, this is the best round up of discussions on genuine multichannel microphone technique to date.
