So I have been reading up on and experimenting with ambisonics a bit and here is something I have been wondering about, are there algorithms (not necessarily for SuperCollider) for dealing with unlimited order? That is I am thinking about something as follows, have an unlimited amount of instruments producing omnidirectional sound, or instruments producing sounds of various solid angles/distances as they move and then having an ability to listen in as these and the listener move around.
I’ll be a little philosophical first: everything is limited. Don’t fall for anything called a “limited edition”, because in the grand scheme of things there is only so much matter in the universe.
Second: unlimited order would require a precision well beyond 64-bit space in a digital realm for scsynth, etc. unlimited means no restriction on bot depth, etc in a digital realm. Infinite.
Computers really don’t like infinites. Really, nature doesn’t either. So you have to find an actual limit to shoot for. Honestly. 3rd is damn good. 5th is rad.
Sorry otherwise. That’s just how our universe is made.
I might have been imprecise in my wording, I guess I mean arbitrary not unlimited. That is being able to change the order depending on the scene and then precisely rendering the scene to binaural within certain sampling rate precision, not necessarily in real-time. I am kind of wondering about what the change in sound would be. Maybe something that simulates the angular sources in the scene and then decodes to headphones/speaker array. That is instead of thinking in terms of order, thinking purely in amount and position of sources. Have techniques for doing this been explored?
There’s always a tradeoff between utility and processing. Higher order ambisonics is a different level of naturalness. Try 7th order using the iem plugins.
There’s also the problem of lossy conversions. Every time you do a math operation, those losses/rounding errors start to pile up. The other thing is, audio is inherently a purely made up construct. Stereo soundstage is something an engineer thought of, so the idea of converting from ambisonics to speaker array for rerecording doesn’t make much sense. On the other hand, supposedly, the easy way of adding reverb to ambisonic signals is converting the output into a T shaped array, but stereo, and applying a stereo reverb to convert back into ambisonic, sounds pretty out there.
What I’d really want would be to be able to place sources in an actual 3D space, and have proper reflections through the space. That way the reverbs wouldn’t be some kind of artificial schroeder algo, but real reverb pushed by amplification of the sources.