[LAU] AMB plugins and Ambisonics implementation

Fons Adriaensen fons at linuxaudio.org
Sun May 13 12:45:21 UTC 2012

On Sun, May 13, 2012 at 01:32:23PM +0200, Ivan Vican wrote:

> My questions: do you know how can I find some Ambdec 16-channel presets?
> I have found several presets in Ambdec directory, but none of them are
> for 16 speakers, so I could really use some help...

There are several 16-speaker presets in the Ambdec distribution, but
they may not be what you need. The decoding matrix depends on the
number and position of the speakers - it is the only part in an
Ambisonics system that does. If you send me the list of your speaker
positions, either X,Y,Z or (preferred) distance, azimuth,elevation,
then I can create an Ambdec preset for your system (free service !).

> Also, I have to say I'm confused with the whole AMB plugins thing (I have
> read papers where people claim they use both AMB plugins and Ambdec during
> Ambisonics reproduction). What is their purpose? Are they just "routing"
> the signals to their corresponding places in Ambdec decoding matrix or do
> they become redundant if you use Ambdec?

Harry has already answered this, but I'll try again using
different words.

If you use a convetional surround system (e.g. 5.1) then you have in
each channel of you mixer a panner that distributes the mono input
signal to a combination of speaker channels so you hear that signal
from a particular direction. The mixing bus is a set of speaker

If you use Ambisonics this is different. There is stil a panner (one
of the AMB plugins) in each channel, but the output of the panner, and
the mixing bus, does not correspond to speaker signals. It is a set
of signals, called B-format, that encodes the direction of each sound
in a way that is completely independent of any particular speaker
positions. You can mix B-format signals just as you can mix a set
a speaker signals. So the mixing bus, and the output of the mixer
is B-format.

The decoder (Ambdec) converts B-format into a set of speaker signals.  

The advantage is that you can take your B-format mix to another studio
or installation that uses a different set of speaker positions, and
nothing in the mix has to change. That other place will use a different
Ambdec preset of course, but that is all. That's why you should not
consider Ambdec as a part of the production itself, it is just a detail
of the playback system.

Another way to look at it is this: the combination of an Ambisonic panner
(plugin) and the decoder (Ambdec) is a conventional panner producing speaker
signals. But since the second part (the decoder) is the same for all channels,
it can be split off and done after the mixing instead of in each channel.
The number of channels in B-format depends on if you use 2-D surround or
3-D (with height), and on the _order_ of the system. Higher order requires
more signals but will provide better reproduction.

As already mentioned, Joern Nettingsmeier's papers are an excellent
introduction to how things are done in practice. If you need more
help, just ask (on the list or privately).



A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)

More information about the Linux-audio-user mailing list