Adventures
in Surround Sound, from 7.2 to Quad
(personal
and historical notes, basics, and acoustic realities often
forgotten)
= P
a r t 5 =
|
Recording
in Surround
Five
Microphones in a Straight Line
Okay,
let's go through the ways one might record a live surround
audio session, as a remote or in the studio. Most surround
recordings will probably be created during the mixing
process, from multitracked sources. That's the same with
most stereophonic recordings, and we'd expect to continue
with new surround mixes and remixes of older albums. But
live recording in the simplest, truly elegant way, tends to
produce the most convincing surround recordings. And the
lessons learned from doing it directly will act as the best
inspiration for what's needed during those elaborate
mixdowns.
The first plan above is
what I think most of us might want to try among our earliest
attempts to record in surround sound. There are five
microphones, one per channel, and they're set in a
straight-line row as you see here. The mikes can be of
several patterns, omnidirectional work well, while cardiods,
even bi-directional microphones can be setup this way with
great effect. If you've ever fooled with a decent variable
pattern mike you know that there's not a huge difference in
the sound quality, but more in the way the ambience and
intensity of instruments gets recorded. In a bright room
you'll probably want a narrower pattern. But in a warm, rich
hall or environment, omni's can be pretty special. I've
shown a KM-86 Neumann unit in these diagrams, as it's an
excellent switchable pattern microphone.
The row of mikes is
situated in front of the performers, near or further away
much as you'd do for stereo sessions. There might be another
group of mikes set up much closer with larger ensembles,
which you'd mix in with these main five to enhance or
"sweeten" a few of the sources, if needed. I'd suggest
trying the ol' kiss principle (keep it simple stupid) for at
least some of your earliest experiments, and adjust the mike
heights, distances and separations, even mike type, to do
the fine tuning. In a classic way the results ought sound
pretty wonderful, if everything else is done
properly.
Five
Microphones in a 180o
Arc
There's
a slightly different approach, which you see here. We can
place the same five microphones along a curved path, perhaps
a full 180 degree arc. The mikes might all be "aimed"
straight ahead (if not omni's), or angled outwards somewhat,
if the performance to be captured is from a large ensemble.
Note that the spacing, as with the first arrangement, is
pretty close to equal spacing between all pairs of adjacent
channels. The reason for trying this variation is inspired
by the ideal way the speakers will be situated on playback.
Since we've seen earlier that a semicircular speaker plan is
hard to beat, it would only make sense to try a similar
configuration while recording. We'll look at a few of the
ramifications down below. Otherwise, what applies for the
first plan will be pertinent for this second version. Early
Stereophonic film soundtracks for Cinerama, CinemaScope and
Todd-AO often used this arrangement, before most stereo
films were mixed into stereo, using panpots and the usual
bag of studio tricks.
A
Common 5-Channel Microphone Array
Another
way those early Stereo film tracks were recorded was more
like this third version. In most cases the rear-side
"surround" track was finally recorded to a single mono
track, so a single microphone placed further away from what
the camera was looking at was common. This works pretty
well, and seems to be a favorite with many five channel
surround recordings of classical music. When the LS and RS
channels are thought of as rear ambience channels, which is
kinda wasteful, but, hey, it's a common notion, the above
microphone positioning seems very logical, indeed.
Then the frontmost
three channels record a typical closer-miked stereophony,
and that's where all the instruments would be heard. The
remaining two channels would be picked up with fairly
distant mike positions, and so would be relegated to
capturing and reproducing the "hall sound", most of the
reverberation and echoes from instruments located up front.
The spacing then might be different from the first two
plans, with equidistant mikes for LF, C and RF, and a much
greater distance to both LS and RS, although they would be
about the same distance from C, and fairly widely
separated.
I think it's mostly a
sad way to throw away most of the intense realism and drama
that five channel surround audio can provide. If those two
channels that cause the most problems, LS and RS are no
longer placed behind the listener, then the matching mikes
need not be so distant, some of the instrumental forces can
be distributed to favor these channels, and the magic of
multichannel can be a lot more exciting. I've drawn the
third plan to match closely the way many surround systems
will be laid out, and for music use, if not film (where the
screen IS up in front, not wrap-around Circarama, fer Pete's
sake!), this can be a very effective way to go,
indeed.
|
(Top
of the Page)
Quadraphonic
Recording
Four
Microphones in a Straight Line
There
are many musicians who've recently commented that a front
center channel is not that important to their music. They've
heard how well two decent channels can create "ghost center"
effects, when sounds are directed equally to both LF and RF,
so see less reason to add more complications to their music
mixes by adding a fifth independent channel. at C. This is
just a revisiting of Quadraphonic Sound, of course. I've
used such a layout for most of my multichannel music, too,
as five channel tape recorders were always rather scarcer
than "poultry dentistry." So what to do but enjoy what you
have, not mourn what you don't have, and may not even need.
Meanwhile the rules are changing again for DVD-A's: many
multichannel choices have become available.
Anyway, I LOVE the
promise of the added C channel, and will not try to
rationalize a four-track maximum here, it's now but one
choice. At the same time, four channel masters can be
"decoded" to extract the in-phase matching level signals
which were placed on LF and RF in the original mix, so that
the home listener will have unique signals on their new C
channel speakers. Many listeners combine their surround
system to serve both music and home theater functions. And
perhaps still a majority of these home systems do NOT have a
discrete center channel, relying on the good old "ghost
center" effect. If the speakers are close together to serve
a video screen, there won't be a big difference between
having a real C speaker or just one of the virtual kind. In
that case we really are back to Quadraphony, but doing it
the right way.
The four-microphones in
a straight line above is what we will probably want to try
for our first four channel surround sessions. We'd probably
also try a curved arc variation, too, as in the second plan
above. Again the mikes can be of many types and patterns,
omni or directional, and the same observations of the first
plan will apply here, too. The mike stands are a little
further apart, to cover the same overall width as before,
but that's about the only difference. And there well be the
same observation to make down below, if the speakers are not
to be placed in a straight line ahead (I've not gone into
this earlier, but it is another option to consider, even if
it compromises the "wrap-around effect" on playback), but
are located in the optimal semicircle.
Creative
Quad Recording -- 2 Spaced Mikes + 1 Coincident
Pair
When
I first began to make remote recordings with that old
Viking
four-track recorder you saw earlier
(and a simultaneous Ampex two track reduction, to play on
the local FM radio station and cut stereo dubs for everyone
else), I found myself often considering the compromises of
the intrusions you might make on a live performance. You
really didn't want to annoy everyone, performers and the
audience, with a maze of tall mike stands, and lots of
cables trailing all over the floor. So it was reasonable to
use the same coincident pair mike setups as we had in our
two-track remote sessions. At the time I was unaware of the
theoretical reasons coincident microphones produce much
better, "tighter" stereo images than spaced pairs can do.
Stanley Lipshitz authored several fascinating AES papers
with audio demonstrations some years back, the
definitive
one in September 1986,
and makes the case very well for NOT using separate, spaced
stereo microphones.
Anyhow, there's an
opposite side of the coin, too, which Lipshitz mentions.
Spaced mikes capture a much better impression of the
spaciousness and sound character of the recording
environment, at the cost of sharply focussed images. That's
why many of the finest recordings of live performances
combine BOTH coincident and spaced pairs. But you have to be
careful, or you can get phase-cancellations and other
disturbing results with multi-microphone pickups
(a
good rule: let one mike per channel dominate somewhat, don't
set them near equal level).
On two channels there are a limited number of choices. If we
record with multitracks more opportunities arise. The first
I thought of is shown above. Trying to avoid a forest of
mike stands, my curious audio friend and mentor, Peter
Downes, and I mounted a couple of cardiod mikes on a single
center stand, then added two extra mikes on smaller stands
to either side. We also tried one of the new all-in-one
stereo mikes on that center stand. Same idea, two channels
from a center position, two from the side mikes. The outer
two could be omni or cardiod, the front pair would be
directional, cardiod or bi-directional (both work well).
Note how it resembles a
three track session to the observer and musicians. And on
playback you hear precise imaging between the LF and RF
speakers, from the coincident pair, while the other channels
are not so sharp in position, but create a marvelous "sense
of place," and spaciousness. It's a lovely way to record in
four channel surround. You can also extract a C channel
later on from the center channels, or use more directional
mikes and locate a third mike element along with LF and RF,
aimed carefully apart. So this plan can be expanded to
include five surround channels.
(No,
I'm not gonna get into the subtle distinctions between
one-axis coincident mikes, versus near-coincident or even
slightly spaced "ORTF and NOS like" configurations -- there
are subtle trade-offs to each. The images here are just for
reference.)
Creative
Quad Recording -- 2 Spaced Coincident
Pairs
On
the other hand, for a major symphony orchestra and chorus
Peter and I recorded for broadcast (the Berlioz' Requiem it
was, instrumentalists and singers located all around a large
cathedral, a perfect subject for surround sound!), this
three stand method was not going to work. There was an aisle
down the middle of the church, and no place to locate a
center stand. No problem, we just shifted over to the above
variation. Now both sides of the large orchestra were
recorded onto four channels from two short, unobtrusive
stands, using TWO coincident mike pairs, a U-48 (facing
forwards) and B&O ribbon (facing more to that side) on
each stand.
(Here's a case that is not directly applicable to five
channels, unless you want to add a single C mike between
these two stands -- hey, that might
work!) You can assume
correctly what the results were: excellent side imaging,
less defined positions in the center, a good wide sense of
spaciousness. What a blast!
It gets even better than
that. We needed more mikes to cover the widely spaced
forces, and the Viking deck had two inputs per channel --
eight of 'em! (Well, it used to sound impressive...) So we
sneaked in a Schoeps cardioid coincident stereo pair on one
tall stand up in front of the main chorus and soloists,
which gave excellent focus and positioning (Peter was as
crazy as me about trying new ideas out). Then for the
auxiliary two choruses and brass ensembles located to the
sides, we hid two more omni U-47's, way off to the left and
right of the cathedral. So those extra mikes were configured
rather like the previous setup above which uses three
stands. Still we were not too intrusive, you had to look
carefully to see the mikestands dotted around. The
theatrical Requiem is composed for major antiphonal forces,
and everyone was being very authentic about this unusual
performance. It was a marvelous recording experience, one
I've never forgotten.
I visited the "scene of
the crime" exactly a week ago by amazing coincidence, the
first time in some decades, and took the above photos of
this beautiful church (as
usual click each for bigger views).
It's the Providence Cathedral of Saints Peter and Paul. The
orchestra and main chorus filled the front spaces
completely, the antiphonal brass sections were placed near
the side altars (a wedding party is rehearsing in the left
one in the photo above) and two additional choruses were
over in each side transept, barely visible here. There's
also a view of the front of the cathedral, with its sturdy
brownstone towers, and the left side facade and entrance. We
setup just inside here, carting everything in through this
doorway, placing the tape machines on a table in the side
passageway. (At the time this plaza consisted of busy city
streets and sidewalks.) We'd been able to assemble four
Ampex 621 powered speakers (amazing sounding devices for
their day), so I was able to monitor with a 120 degree arc
of speakers, while Peter and his wife, Maggie, were singing
in the main chorus for the Berlioz.
We used a similar setup
several times again during my final years at college, and
also when I had moved to NYC for graduate school, when I
could get away to help Peter with other thorny remote
recording sessions. The final remote session we did together
was an even more elaborate quad session in the Atlantic City
Convention Hall Auditorium (wotta huge barn!), but I've
rambled on enough here already -- that's another story for
another time...
|
(Top
of the Page)
Multichannel
Recording Quirks
Recording
"Depth" Quad (diamond
configuration)
Early
in this web resource you'll find reference to a
long-forgotten suggestion of how to use four tracks for much
greater stereo realism, which
we've called "Depth Quad."
We saw how such a recording could be played back, with a
"diamond" arrangement of speakers, all of them up in front,
adding a center close
and a center distant
channel to a wider than usual L and R stereo pair. Let's
view the setup to record such a master. It's pretty
straightforward. The mikes are in essentially the same
locations as the speakers will be. They will probably be
spaced further apart, to "scale" up to a larger performing
space than our home playback room is likely to be. No
matter, it's a one-to-one correspondence.
I realize it might look
a little silly here, when I threw in the chair from my
studio again to give it scale (ick, the perspective doesn't
quite match, nevermind...). But smile as you might at
sitting with a mike or speaker in your face, you ought try
it out for yourself sometime. After all, we won't see any
commercial recordings available using this idea anytime soon
;^)... you'll have to roll your own. It reproduces an
uncanny sense of depth, much better than anything we're used
to, or about to get used to. But there are tradeoffs as well
(not much overall width, and no wrap-around, depth-cues
confined mainly to the center). I'm not actually
recommending this as our new multichannel system. We'd need
at least eight channels to do a convincing job with depth,
as described before. And THAT many channels does seem to be
premature for the moment. Keep it in mind, though, for a
future stage of audio evolution...
Performers
and 5 Microphones in Straight Lines
Every
good idea has its down side, too. Above we looked at some
microphone arrangements that are simple and effective. I'll
stand by my suggestions for the optimum final playback
arrangement: five channels in a deep 180 degree arc. Still,
we might go for the straight line arrangement to record a
live performance. It's simple and presents a less cluttered
appearance to the performers and audience. Or try one of the
alternates described above, with fewer mike stands. Fine,
here we have recorded a hypothetical live concert, a small
chorus with piano accompaniment, something you might very
well get a chance to record in your hometown. Everything
goes according to plan, and we bring the master back to our
studio (you may guess where this is heading). Start the
playback, and here's what we hear:
Reproduces
with Illusory Curved Arc Configuration
Suddenly
what emerges nice and cleanly out of the monitors is what
you see here: a chorus all around in a semicircle, the piano
up front and center. It's dramatic and makes for an exciting
listening experience. BUT -- it's not what you began with.
The relocated five channels have "warped" the soundfield
from straight across the front to a wraparound virtual
chorus, depicted here. If you go the other way around:
curved mike locations (the second diagram above), playback
with five speakers in a straight line, the opposite
"warping" will take place. To maintain the original
configuration of the ensemble, you have to match plans for
both mikes and speakers. And in this case, I think you'd
hear that a deeply curved pickup of a straight across
performance would tend to waste some of the realism of five
channel surround, as the LS and RS channels wouldn't be
doing their fair share. If surround music is to catch on
with the public, let's show it off properly!
Anyway, it is called
the recording arts
and sciences, isn't it? There's no reason to apologize for
the creative side of the act of recording, what effect you
must have on the results. There is also no such thing as
100% accurate reproduction, and never will be. If the final
product is good fun and an absorbing listening experience,
if it conjures up an idealized "real" performance, expanding
your own listening environment, who cares? My use of
"warping" above is mostly a self-effacing tease. Don't fret
the lack of exact match between "whatguzzinta" and
"whatcumzouta." The goal is to create the
illusion
of life, the
illusion that everything's identical, and just like making
films or animation, you use whatever artifice it takes
(those being very artificial art forms).
The reason the speakers
ought be situated in one of these simple 180 degree curves
is simply to present the maximum audible directional clues,
with least wasted or ambiguous information, to the human
hearing apparatus. Sometimes you may actually capture nearly
what occurred in front of the mikes, and the listeners will
hear it that way (depth quad does that in a limited way, and
so can binaural sound). Most of the time it will be more
complicated than 1:1.
But isn't this what cutting edge recording OUGHT be all
about, "creating" and "recreating" all mixed-up together,
producing wonderful listening experiences that can't be
obtained in any other way? Everyone heard the improvements
of stereo over mono sound in the 50's and 60's (and two
track stereo is very artificial). Everyone does NOT hear the
purported "angels on the head of a pin" improvements of
gross oversampling and redundant data wasting. Properly
done, there will be equally obvious "impact" as stereo had,
when we advance to multichannel surround sound. Stick around
and join the adventure!
--Wendy
Carlos
©
Copyright 2001 Wendy Carlos -- All Rights Reserved
|
(Top
of the Page)
Back to the Wendy Carlos Home Page