conceptual ambisonic audio question
This isn't Max/MSP specific, but I thought this community more than others that I'm connected to would be able to help.
When dealing with stereo audio, panning essentially "moves" the signal strength completely to one side or the other so that a hard L pan will make the entire signal heard out of the L earbud/speaker and silence would occur out of the R earbud/speaker.
I want to replicate that concept with audio encoded into a YouTube 360 video. Say that the camera is located in the center of a room with four walls, and each wall is playing a different video. I want the listener to be able to rotate around the 360 space and have the sound of the wall that they're facing be treated as a "hard pan" - they hear the audio of that wall and all of the other walls are silent and/or very distant.
I've experimented with this and also done a bunch of research and it seems like this might not be possible - I know that I can create a binaural audio file attached to the video that can make the current facing louder, but I can't seem to find a way to attenuate the signal of everything else around to silence, or at least to the point where it is incredibly faint background. I feel like the nature and approach of binaural ambisonics is not designed for this kind of thing since the point of immersive audio is to have audio in an ambisonic sphere be as realistic as possible and this breaks that.
Does anyone have any insight as to how I could potentially achieve this, or am I trying to fit a square peg into a circle hole?
how many audio sources do you have? 1 * video/wall = 4 (if stereo=8)
via logic mute gradually the other 3 while point at 1
It is true that ambisonics reproduces a sound field, but you still can have control on individual sources gains
but in a YouTube 360 video? i don't think i have the ability to dynamically change that based on how the viewer wants to view the material. Can a gain "map" be created for something like that based on the facing of the viewer?
Hi Mendel, I also think you can't make it dynamically, since you are uploading a fixed sound filed in B-format AmbiX or whatever.... maybe you could try with 4 mono sources from the walls with less bass/difuse content?
Regarding to the binaural render for a video you actually can process the ambisonic sound field this way, for instance check ATK (SC or Reaper), I'm not aware of any tool like this in Max at the moment but sure there is something similar.
many thanks Juan! I'll check out the ATK and just keep experimenting.
Cheers!
Sorry my bad, I thought you were after something interactive, but afaik not really possible, since it is true that youtube contemplates the rotation and elevation of the head, but not the movement in the field, which could have been useful, because positioning sources far away from each other would have probably helped.
in utube 360 you deliver a 1st order ambisonic RENDERED file (actually it is ambiX format where the order of channels is inverted, W YZX) . It is the rendering engine/system which then decides on how to reproduce/decode the data (this is the main reason why to use ambisonics -> pre decoding)
So the binaural file (which is a 2 channels file) is not what you are after, although one needs to monitor, which is another thing.
please check very well on how youtube manages this. there is also a metadata app which you should check out.-
https://github.com/google/spatial-media/releases/tag/v2.1
and this guide, simple but useful
http://www.bili-project.org/wp-content/uploads/2016/12/HD-LH-TUTORIAL-Youtube360.pdf
ATK -> wonderful, and reaper is becoming a very efficient DAW for multichannel.
For binaural monitoring you could also give a try to this, not really working yet on windows (unless you can compile several libs)
https://zenodo.org/record/3269271#.XXYOICgzaUk
people at huddersfield are great
Thanks so much Jaky! This is probably good incentive for me to finally pull the trigger on buying Reaper which has been on my radar for a while.
Cheers!