View Single Post
Old 07-25-2016, 01:47 PM   #6
lossius
Human being with feelings
 
Join Date: Jul 2011
Location: Bergen, Norway
Posts: 123
Default

Quote:
Originally Posted by ReaDave View Post
Regarding the pseudo inverse situation, do you have any information about what that does and how not including it will affect my listening tests? I didn't see much of a description about that in the tracker.
I have been directly comparing your cube decoder with Bruce Wiggins one and, from the limited time I've had so far and my less than optimal current speaker configuration, I prefer the sound of yours. It seems to provide better rear localization. I did have to reverse the polarity of the two lower rear speakers though for the sound localization to work well enough. As I mentioned in my other thread though, that could be either an amplifier issue (I'm running those speakers from a different amplifier) or it could be because those two speakers are not Auratones and I've had to EQ them to get them timbrally similar.

Or perhaps this is related to the pseudo inverse thing you're speaking of?
The PseudoInverse thing might not be clear from the tracker, but at least I myself understand what it's about when looking at it... ;-)

The idea with ambisonics decoding is that it is very straight forward when all speakers are positioned at equal angular distance. In 3D this only happens if the speakers are laid out according to one of the platonic solids. My guess is that Brice Wiggins Cube decoder is using one of these shapes, and if so, the vertical distance between upper and lower ring has to be the exact same as the distance between front and back speakers in order for the decoder to work properly. This might be why you are not getting satisfactory results from it.

If you need to decode to irregular speaker layouts, some kind of "best fit" solution need to be applied. The PseudoInverse trick is that it asks what it would sound like if you had sound sources coming from the direction of each of your speaker. It then inverts this so that what started out as an encoder of a number of planewaves instead becomes a decoder. I´m not sure exactly how that will sound and how it will differ fro the current implementation, but I trust Joseph when he says that it will improve the decoder. But that being said, I agree with you that it works fairly well already, I have been surprised by how good it sounds when trying out with two rings of 8 speakers at LABRI in Bordeaux in May. And I'm looking forward to test with proper PseudoInverse implementation as well.
lossius is offline   Reply With Quote