[LAU] Ambisonics: Which order fits my project?

Fons Adriaensen fons at linuxaudio.org
Sun Jul 5 12:48:00 CEST 2020

On Sun, Jul 05, 2020 at 10:31:51AM +0200, Michael Jarosch wrote:

> Speaking about Ambisonics, I have one question regarding the different
> orders you might want to use, which I couldn't found to be answered in
> the (not too much) media I read and watched about this topic:
> Which border do I have to reach in my project that I should consider
> switching to higher (second, third or more) order ambisonics? Is it the
> number of output channels? Or is it: Always use the highest order you
> can afford?

The first question should be if your project would benefit from
Ambisonics or not. I have been using and advocating Ambisonics for
a long time, but I can easily imagine situations where it would
be the worst way to do things.

AMB makes sense if

1. you not only want to put discrete sounds in the right direction,
   but also create virtual space that sounds natural,
2. there should be no 'preferred directions', 
3. you don't require effects that happen only in a specific
   place, e.g. close to a sound source,
4. you need operations like rotating the entire sound field, or
   many moving sounds,
5. you can distribute the speakers more or less evenly in all
6. you have a method to create the content in AMB format.
> According to this video <https://www.youtube.com/watch?v=f4oE3i2tjyA>(,
> which I haven't watched the whole length,) at 23:14 ff it's about
> frequency response and compensation - and there it seems to be better
> using lower order.

That is all related to obtaining HOA signals from spherical microphone
arrays - the higher order components will have a restricted frequency 
range. None of this applies to AMB as a rendering system.
If creating the content for your project requires the use of such
microphones, then this may matter. Otherwise not.
> It would be interesting making a direct comparison between first and
> third order to hear if it's worthwhile.

It certainly does. Above third, there are still benifits, but mainly
in bigger spaces. Another big difference, if you reproduce natural
sounds or recreate the acoustics of some space, is between 2D and 3D.
Adding the height dimension helps a lot in such cases, even when
using lower orders. But it comes at a price: the number of channels
and speakers increases as the square of order.

In many cases, (6) above may well be the deciding criterion. The
only way to produce higher (3rd and above) order AMB (without the
limitations of spherical mic arrays) is to mix and pan individual



More information about the Linux-audio-user mailing list