[ardour-dev] quick note re: MTC

[Fons Adriaensen]

On Tue, Jan 02, 2007 at 03:08:35PM -0500, Thomas Vecchione wrote:

> > But you can still assign channels to the
> > 5.1 signals after the decoder of course, for example movie dialogue
> > to the center speaker.
> 
> 
> Which I hope is kept in mind when the panner is rebuilt if it does use 
> Ambisonics, as there are situations that do require it to only come out 
> of a single speaker. 

Ardour is flexible enough to do this. The type of panner you get
will depend on the number of outputs you give to any particular
mixer channel:
 
1 -> no panning,
2 -> stereo pannning, as it is now,
3 -> first order AMB panner, horizontal only,
4 -> first order AMB panner with height,
5 -> second order AMB panner, horizontal only,
6 -> second order AMB panner, with first order height,
7 -> third order AMB panner, horizontal only,
8 -> third order AMB panner with first order height,
9 -> full second order AMB panner.

There should also be the option of having no panner at all, for
any number of outputs.

For example for ITU 5.x you would use a 5 channel mixing bus,
and most mixer channels would have a second order horizontal only
panner. The corresponding bus module would have a AMB->ITU decoder
plugin, feeding a six channel speakers bus. There is nothing to stop
you having a mixer channel that feeds directly to one or more channels
of the speaker bus - Ardour can do this easily.

> Out of curiosity, to ensure I understand Ambisonics correctly, in a 
> panner implementation of it, you would end up essentially putting in a 
> scale and placing your speakers in the places they will be in the end 
> result correct?  So if I have a 30 Foot stage, with a center cluster 
> above centerline, and two stereo clusters 20 feet off center on either 
> side, and 6 speaker positions placed throughout the space, I could 
> essentially place those correctly in the panner and it would take care 
> of appropriate delays etc needed for the panning as well(For instance if 
> the sound is coming from the Stage right area, panned in Ambisonics, the 
> opposing speaker in the house may produce some sound, but it would be 
> delayed so the sound from the stage would reach there first?)

It would take much more than this post to explain all of Ambisonics, but
there are two important points to understand here:

- The idea to represent the direction of a sound in the way it is done
in Ambisonics (i.e. using an Ambisonic panner) has nothing at all to do
with the rest of Ambisonics technology. It is just the 'mathematically
obvious' way to encode directional information, more or less in the
same way as complex numbers map naturally to points in a plane, or HSV 
is a 'natural' way to represent colors. The amount of 'detail' increases
with the order of the system, which translates directly to the number
of bus signals needed.

- The essential point here is that the way the directional information
is encoded in the signal (in the example, in the five bus signals), is
independent of how it will be rendered - the bus signals are not speaker
signals, and they have to pass through a decoder before they can be sent
to a speaker. You could replace the ITU 5.1 decoder by another one that
feeds for example 6 speakers in a regular hexagon (giving much better
results than 5.1), or any other layout. All the necessary processing
(including e.g. delay) is done by the decoder, not by the panner. The
panner is completely agnostic of the reproduction system. Which means
for example that if you distribute the signals as they are on the
mixing bus fed by the panners (i.e. before the decoder wich you use
to listen to the result), they can be used for reproduction on any
(sensible) layout of speakers.

Ambisonics - in particular higher order - is the best technology for
creating a 'you are there' effect, e.g. for natural sounds or for 
classical music where it can capture the entire acoustics of a concert
hall and allow this to be reproduced exactly. It is not necessarily the
best technology for theatre sound, but it can easily be combined with
other methods.

--  
FA

Lascia la spina, cogli la rosa.