[LAU] Jconv

Fons Adriaensen fons at kokkinizita.net
Sun Jan 11 07:57:55 EST 2009


On Sun, Jan 11, 2009 at 12:56:38PM +0300, alex stone wrote:

> Fons, just a follow up to the discussion about an orchestral placement
> template, is there a definitive resource somewhere i can go to, and learn
> more about ambisonics, and how this might help me to achieve the Holy Grail
> of true placement recording?

The simple answer to this is 'no' :-(

You can google Ambisonics of course, and you'll find some
sites. Then you'll find out that information is fragmented,
often contradictory, frequently outdated, and in some cases
plain wrong. One of the reasons is that the technology is
now well over 30 years old, at the time it was invented it
was only possible to use it in its simplest forms which
requires many compromises deviating from the theory, and
some of these compromises have over the years established
themselves as 'the thruth', confusing anyone who tries to
understand things from a theoretical perspective. And of
course at that time we didn't have the web. The last ten
years or so have seen a renaissance of AMB research, and
of course today many of the technical barriers have gone.
Some people are now using up to 4th order systems routinely,
and some are researching much higher orders. Today the AMB
community seems to be divided in two camps, the old guards
that were part of the original wave in the 70's, and a
new generation of researchers that almost restarted from
scratch, and is not bothered by the same legacy wisdom.
Their publications (mostly PhD theses and AES papers)
today provide the best information, but their approach
is highly mathematical, something that can't be avoided
except for the most simple cases. In between those two
is part of the electro-acoustic music community which
has been developing its own toolsets, most of which are
incompatible with each other. I've been involved during
the last half year in some 'AMB standards wars', and it
has not been a very happy experience. 

All this should not stop you from learning about AMB, but
you will find conflicting info and have to sort out the
mess - you have been warned :-).

First question of course is if you need it or not. If your
end result is high quality surround or periphonic reproduc-
tion you definitely do. If it is stereo you don't really
need it, but even in that case an AMB based production chain
can do some things that would be difficult to achieve in 
any other way.

Ambisonics can be understood on many different levels,
from 'intuitive' to purely mathematical. On the intuitive
level, it's a way of representing the directional (not
distance) distribution of sound in a 'canonical' way that
is really independent of the way the sound it is captured
or will be reproduced. In other words, AMB signals are not
speaker feeds, but a something that can be 'decoded' into
speaker signals. In its simplest form you could compare
it to representing colour video not as RGB, but as 
(intensity, colour, saturation) which can be decoded
to RGB, or to any other colour representation.

On a first mathematical level, you can see this as a sort
of 'spectral' representation. Any cyclic function can be
Fourier transformed into a set of harmonic frequencies,
each having its level and phase. In similar way, the
horizontal distribution of sound directions is a cyclic 
function, not of time but of the horizontal angle (azimuth)
of the sources, and you can apply the same Fourier transform
to it, which is how horizontal AMB works. The 'order' of an
AMB system refers to how many 'harmonics' are used.
For a periphonic (3D) sound distribution the 'function on
a circle' becomes a 'function on the sphere', depending on
two variables, azimuth and elevation. Because a sphere is
not the same as a 2D hyperplane the corresponding spectral
transform is not the 2D FT, but is defined by the set of
'spherical harmonics'. And here of course the more difficult
maths start...


> I've read back through your comments about
> routing to A and B to achieve 'local' and overall
> placement, re convolution, but i will admit i don;t
> understand all of it.


I'll comment on this later.

Ciao,

-- 
FA

Laboratorio di Acustica ed Elettroacustica
Parma, Italia

O tu, che porte, correndo si ?
E guerra e morte !



More information about the Linux-audio-user mailing list