Having read some of the discussions on reproducing
an ambient sound field, it might be useful to
explore some of the theoretical issues. All
technical solutions are crude compromises.
Although Chapter 6 of the book, =93Spaces Speak,
Are Your Listening?=94 explores the issues for music,
the same situation presents itself with all natural
sound field recording.
The central problem is that one is trying to
reproduce sound in a volume of space.
Microphones and microphone capture and reproduce
sound at a few discrete points. Mapping discrete
points into spatial volumes is THE problem.
In this sense, the compromise is always in the size
of the sweet spot, the volume in which the
reproduction is sufficiently accurate. At on
extreme, we have the volume at the entrance to the
two ears, which approximates discrete points in space.
At the other extreme, we have a room where many
people can be in a large sweet spot. In this case,
a perfect reproduction would allow a listener to walk
through the reproduction space as if he/she were
walking through the original space. Each of the two
extremes presents two different classes of problems.
Take the large sweet-spot first. Perfect reproduction
requires the sound field to be sampled in time AND
space. Based on the Nyquist theorem, one has to same
at twice the highest bandwidth. Assume 10 kHz for
discussion. But one also has to sample in space.
The wavelength of 10 kHz is about 0.1 inches.
Therefore a 1 cubic foot volume requires about 10**9
points to be sampled. You get the idea. The numbers
become so large, that perfect reproduction is clearly
impossible in a large volume. Moreover, the reproduced
sound field takes place in a space that has its own
acoustics. And each listening space is unique.
Now take the other extreme, which is binaural recording.
The sweet spot covers only one individual. But that too
has two issues. An individual=92s outer ears, the pinna,
are unique and the recording process using a dummy head
needs to have ears that match those of the listener.
Moreover, with headphones, the sound field moves as
the listener turns his head. Once we introduce a moving
head, we are enlarging the volume in which the sound
field needs to be captured and reproduced.
Having established that a perfect production is
impossible, we are now left with the question of how
to choose a compromise.
One obvious choice is to borrow the commercial technology
of 5.1 home-theater. Another choice is to borrow some
of the more advanced laboratory systems, such as
ambisonics, holophony, or head-tracking binaural.
In conclusion, the critical question is deciding the size
of the sweet-spot. Is the reproduction intended for one
listener or many? Is the reproduction equipment part of
the recording assumptions such that a listener must setup
a particular system in order to listen to the recording?
Make your compromises but be clear about their assumptions.
Even the best systems are crude compromises.
|