[Gary Eickmeier]

"Doug McDonald" wrote in message
...
Actually I suspect that 4 mikes is enough to establish a
soundfield and am quite sure that 16 would be.


That is for the soundfield impinging on one spot of course.

The four mikes would be four omnis in a spaced tetrahedron,
say 10 inches on a side.

The sixteen would be in groups of four coincident ones at the points of
the same tetrahedron.
These would consist of one omni and three figure 8 ones, the three figure
8 ones being pointed up-down, east-west and north-south.
Or, alternatively, four cardioids pointing out from the center
of the tetrahedron.

This would allow computerized localization of
a single sound source generating a sine wave, or
localization of the original and reflections of a
point impulse source. Of course doing this for a whole
orchestra would be an immense and probably somewhat
impractical computer programming job. The accuracy would
decrease at low frequencies of course.

OK as long as we are stretching minds here, permit me to do another thought
experiment - which could easily be a real experiment if we had time etc.

We will record a singer with one mike, a drum kit with another, a piano and
perhaps a bass, all with their own microphones. They are playing in a good
hall or studio. While recording the piano on the left side of the room we
also catch the early reflected sound from front and left side wall.
Similarly for all the other instruments and voice. The recording now
contains information about both the direct sound of that performer and some
of the reverberant from near them.

We will play back each channel to a fairly omnidirectional speaker placed in
a position similar to the performer it is reproducing. The singer, for
example, will be a sound image that is nailed in the center of your room and
pulled out from the front wall of the room. You can now walk all around and
magically hear the singer where she belongs, three dimensionally singing
right there in front of you. Any problems so far? We recorded her with but
one mike, paid no attention whatsoever to the human hearing mechanism, HRTF,
spacing of ears, dummy heads, none of it. When we listen to her on playback,
we each use our own human hearing mechanism and HRTF and all that to hear
her, each in our own way in our own head, so there is no question of
worrying about any of that.

The piano will be played from a speaker or speakers on the left side, and if
it is basically omnidirectional it will cast some reflected sound toward the
front and left side walls just as the live piano did, and you recorded. The
spatial impression we will all get from this arrangement on playback will be
the same for all of us and very similar to the live event, and again we can
walk all around and the piano will remain nailed right there where it
belongs.

Similarly for all instruments, the whole system becoming more like a "model"
of the live event than a "picture" taken from one spot in the studio with a
fixed perspective. Is this getting spooky or what? Stay with me.

What we have here is a field-type system, a system which was recorded and
reproduced with no reference whatsoever to the human hearing mechanism or
the position of a typical listener to the live event. We play it back by
positioning the speakers in another acoustic space, in a certain way that
models the playback situation after the original, and we have a "stereo"
phonic, meaning "solid" sounding reproduction of the original.

If the original was played in a more live acoustic, like a church or a cave
or a good concert hall, we would have preserved most of the temporal aspect
of that reverberation, and on playback it is up to us to recostruct the
spatial aspect with correct radiation patterns and positioning of the
speakers and enhancent from some surrond speakers, fed by either additional
mikes or signal processing. If we take that approach, we get a pretty good
impression of the sound of the original event, even though the temporal
aspect of our playback room is superimposed on the recording a little.

But wait - what if we use quite directional speakers on the same recording?
Well, this would be a big mistake, because then all of the sound, the direct
and the reverberation, would be forced to come from the location of that
player, an unnatural sounding situation. The piano, for example, would have
all of its output coming from the speaker box rather than from both the
speaker and the left side wall. It would be more "accurate" if you compare
the wrong aspects to each other - the electrical input signal would be JUST
LIKE the acoustical output of the thing - but the resultant sound would be
nothing like the original - accurate but not realistic! What a mess.

This experiment is the paradigm for the field-type system that we call
"stereophonic sound" in general, presented to you as a basis for
understanding the simplification of it that we call - well, the same word is
unfortunately used. The extremely fortunate psychoacoustic principle that
permits us to simplify this more elaborate model down to fewer speakers and
microphones is called summing localization. This is the idea that you can
image anywhere between two coherent sources by means of intensity or time
difference or a combination of the two, permitting the impression of many
instruments from just two or more mikes and speakers.

So the "big picture" of what we are doing with "stereo" is something like
close miking the soundstage rather than each individual instrument, but the
principle remains the same. We can get away with reproducing an entire
symphony orchestra with just two speakers if we position them in a certain
way and let them reflect a certain amount of their output from the
appropriate wall of the playback room etc etc in a way that models the
playback after the (typical) live event.

This grand simplification can be engineered with time/intensity trading so
that you can walk across the room and have the imaging remain solid and hear
them from some degree of variable perspective. You can also engineer the
directivity index of the speakers so that the RANGE OF RATIOS of direct to
reflected sound is similar to that in the (typical) concert hall with the
very much greater distance of the listener from the source.

As I said, there is very much more to study about all of this, but we will
get nowhere until we can understand the basic principle and how it differs
from what we have been doing (or not paying any attention to) for so many
years.

It's a whole deal.

Gary Eickmeier