Here's a recording excerpt that ties together some past discussions
of MS with recent discussions about realism/impressionism, and
includes a bit of fieldcraft and post-processing along the way.
This recording of frogs calling on either side of a waterfall was
made in Borneo last November. Perhaps give the recording a listen and
judge it on its merits before reading on...
http://simmosonic.googlepages.com/hornedfrogs.m3u
According to Martyn Stewart - our tireless moderator whose initials
are also a wonderful stereo microphone technique - these are horned
frogs (megophrys nasuta). Martyn kindly identified them for me in a
post here about two months ago, back in the days before he curiously
forgot that I do actually make nature recordings.
Walking at night with my MS rig, I heard the frogs calling and, after
a bit of careful treading off the path and through marshy ground,
managed to place the microphones in a position that put one caller
hard left and the other hard right to create an interesting
left/right call/response effect.
There was a nearby waterfall that I wanted to place in the centre of
the image, creating the impression that the frogs were calling from
either side of it (which they were, of course, if you found the
correct vantage point). To do this I had to choose a diagonal
position with respect to the frogs, which resulted in the right side
caller being quite close, and the left side caller being relatively
distant. I didn't mind this because such things can add a sense of
depth. Besides, I couldn't ask the left caller to move closer;
languages are one of my weak points and I don't speak megophrys
nasuta very well.
Listening back to the recording in the cold light of day (i.e. the
next morning, when I could be more objective), I realised that the
waterfall was too far away to be heard clearly and was really just a
lot of muffled low frequency rumbling; it was spectrally out of
balance with the frogs. So that night I returned to make a recording
of the waterfall to blend in with the frogs. For this, I moved much
closer to the waterfall, switching my monitoring between the frog
recording and the microphones and moving in and out until I thought
the spectral content of the waterfall matched that of the frog
recording. I suspected the image was too wide to match the width of
the waterfall in the frog recording, but I knew I could fix that
later by adjusting the MS ratio.
An alternative approach would've been to record the waterfall at a
slightly further distance, to get a narrower image, and then use some
EQ to match the tone to the frogs, but I preferred to match the tone
in situ. This was the more 'pure' of the two approaches, IMO, and
also easier to judge in headphones. Besides, the whole point of
returning to the waterfall was to capture a more detailed sound, so
closer was my preference.
Both of these recordings were made in MS, using the same microphones
and recorder. This makes it easier to match the recordings together,
tonally.
When overlaying the two recordings, I realised that I needed to
remove the muffled rumbling of the waterfall from the frog recording -
it was adding a lot of unwanted LF information that was affecting
the clarity and detail of the frogs due to low frequency masking.
Assuming that most of the muffled rumbling was in the M channel, I
applied the high pass filter to this channel only at first, prior to
decoding to stereo, hoping I could leave the frog sounds (dominant in
the S channel) untouched. But the waterfall had significant low
frequency energy in the S channel as well, so I decided to pass both
signals through the same high pass filter. (I could've passed both
channels through separate high pass filters prior to decoding to
stereo, optimising each one's cut-off frequency and slope to get the
best effect with minimal effect on the S channel, but that wasn't
necessary.)
Then I reduced the level of the M channel ever so slightly. Although
this widened the frog recording a little, it did not have an adverse
effect. The goal was to create a tonal hole in the middle of the
recording to place the waterfall in. I now had a relatively clean and
detailed frog recording, awaiting the return of the waterfall.
A similar high pass filter was used on both channels of the waterfall
recording to produce a sound that I felt was more how we perceived
it, rather than what existed in reality (which tended to overpower
the frogs).
As suspected during the recording, I had to reduce the width of the
close-up waterfall to match the width of the waterfall in the frog
recording; otherwise, it was too wide (relative to the frogs) and was
obviously an added feature.
Some might call this a 'composed recording', but if I had four
recording tracks and two MS pairs, I could've recorded the frogs and
waterfall simultaneously and done similar processing to achieve the
same or similar result. Would it still be composed, or simply a multi-
miked recording? And what if I had mixed and recorded all four mics
direct to stereo?
Sometimes I play this recording and am very happy with it; other
times I think the waterfall is too up front, or not quite loud enough
to be detailed, or any one of numerous other criticisms.
I'm not presenting this recording as a piece of art or example of
good recording (the waterfall actually hides a lot of problems in the
frog recording!), but hopefully some readers might enjoy reading
about the process of creating it, and the use of MS processing after
the recording. It has found an application on line for which I have
been rewarded for, so I have no complaints...
- Greg Simmons
|