editor's note 2007
This article appeared in Wildlife Sound March 1973 ff. Modern listeners may not appreciate just how new stereo recording was at the time. Decca had only started recoding music in stereo in May 1955, reputedly the first European recording company to do so. It is a tribute to David's thoroughness that the vast majority of this article is still relevant, barring the section on recorders which has been overtaken by digital recording in the mid 1980s.
Bearing in mind this was written when stereo field recording was in its infancy it is surprising that only one microphone technique which is now common in wildlife and field recording today is omitted, the use of Mid-Side pairs.
The microphones used in the series of experiments conducted were of the very highest standard in broadcasting terms, of a highly predictable and well defined performance which satisfies the professional and should truly delight the amateur. Having this advantage of predictable performance microphones, the task of using them to obtain a similarly high and predictable standard of stereo on wildlife was, I think, made a little easier than it would have been for amateurs operating with perhaps limited resources and knowledge.
If your are to obtain good stereo by experimentation, I think the first thing to do is to become attuned to stereo: by this I mean listening to good stereo for a few hundred hours to train the ears and mind to accept stereo as the normal medium of listening. After this long period of stereo listening, mono sound will be worse than just dull, always assuming you have two good ears.
Remember stereo is not natural sound and it should be limited to sounds coming from a fixed angle, not from two points but between the fixed points. Stereo should, however, give the listener a degree of realism, which nearly matches a listener's vision. Sight is limited to a forward angle which gives the viewer, with full lateral eye movement, about 75 degrees forward width, and for some odd reason this seems to be the ideal angle for the listener to be in relation to the reproducing loudspeakers.
When fully attuned to stereo you will then be in a position to experiment and should be able to spot many of the problems associated with wildlife recording in stereo. One of the most likely faults to be experienced first is the hole in the middle, i.e. lots of sound emitting from both loudspeakers, with only a little appearing to come from between them, whereas, in fact, just a little sound should appear to come out of each loudspeaker with most of the information apparently originating from points between them.
Anyway, as I have said, the first essential is to know how to listen, and this must be done on loudspeakers which are of matched, or nearly matched, performance. Headphones are of little use except for field work monitoring. I cannot emphasize enough that hour upon hour of good stereo listening is absolutely vital before one starts to experiment. Only on this firm basis will it be possible to produce good stereo yourself.
The problem of long-range recording can perhaps be likened to vision again. If we wish to see distant things clearly, we limit the angle and increase the magnification by the use of binoculars. The angle presented to the eye should still be in the order of 70 or so when using good binoculars. Likewise we will wish to present a good angle to the listener of these distant sounds. The microphone does not have a sharply defined angle of acceptance like the optics of binoculars, particularly in respect of the lower frequencies in the audio range. The use of rifle or gun type microphones would give too great an angle for long range work, so the only device which can be employed is the well proven (in mono) parabolic reflector. The parabolic reflector's performance and shortcomings are well known; the forward low frequency performance is limited by the diameter of the reflector, thus the low frequencies tend to be picked up equally from all directions if not controlled by a very large reflector; i.e. around 10ft diameter. The middle frequencies are of fairly wide angle and the high frequencies very narrow, assuming that the microphone is set at the correct point of focus on the axis of the reflector. The forward gain is low at low frequencies, high at middle and high frequencies; but, as already mentions, the pick-up angle at middle frequencies is fairly wide, much wider than at high frequencies, hence the quantity of sound picked up is a lot greater at middle than at high frequencies.
The reflector is therefore not a high quality device for picking up the majority of sounds, particularly atmospheres, but is acceptable if used on single subjects when the background sound is low. Although the reflector is not of high sound quality, it does do the job of gathering the required distant sounds.
Experiments were performed with two 36" parabolic reflectors fitted with both omnidirectional and cardioid high quality microphones.
The two units were placed close together side by side and angled from zero to 15° outwards. The focus point of the microphones were also varied from mid-distance through to long distance.
As a result of using these various combination the facotrs emerging were as follows: no stereo in the low frequencies; reasonable stereo in the 500Hz to 2kHz range. In the range above 2kHz, if the angle increased more than 5 (which was necessary in order to get any stereo in the 500Hz to 2kHz range), a hole occurred in the middle.
This time the previous tests were carried out with the distance between the reflectors increased in stages up to 30 paces (or yards), and the axes from the straight ahead position to 30 crossed.
The general effect was quite impressive, but the main faults were
The use of two reflectors for stereo produces poor results
Having heard about a divided stereo parabolic reflector used by Sten Wahlstrom (See introduction), I decided to convert a standard 36" unit. A ¼" plywood acoustic screen or shield was fitted vertically across the dish splitting it effectively into two half sections.
After much experimenting, the microphones were placed at the long distance focus inclined inwards to face the reflecting board. Completely felting he board, and other variations, were tried, but the best results were obtained with the plain board.
Stereophonically, the background was good and the subject movement of 10 either side of the reflector axis gave substantial stereophonic movement. This is a narrow angle device and the pick-up intensity falls of rapidly 10 off axis. There seem to be some phase problems with sound coming from angles 10 or so vertically above the axis. It was also noted that where calls have both high and low frequencies, the reproduced low frequencies did mot come fro the same place as the high, i.e. if the subject calls left (high frequency) then the low frequency swing to the right on a subject position 5 or more off centre axis
As relatively few wildlife subjects have vocalisations containing low frequencies, the low frequency shift is of little concern. Therefore, the divided reflector is, so far, the only device worth using for long-range recording in stereo. Indeed, the results are very worthwhile once you have discovered the correct position for them microphone.
Web editor's note 2007
A modern practice for getting stereo from a parabolic dish is a slight variation on David's preferred technique. The baffle between the microphones is around 4-10cm large rather than a full separator dividing the dish. See the article Parabolic Stereo for a more detailed discussion of this technique, and the article MS with reflector for a slightly different approach.
In part 2 I intend to deal with the recording at medium ranges (both narrow angle and wide angle) and close range (narrow and wide angle) plus information on microphone phasing and angles.