< Back to IRCAM Forum

Other set up and directivity

Spat
#1

hi T,

i m lucky to work on a very nice set up made of 45 speakers. But unlike the ones i worked on so far, these ones are set up as a grid/cube, not spherical, and multi layered (speakers above other and bellow), which is cool too. I’d like to see where esthetically it could bring. I m thinking of trying spat at some level here however they also have their own audio engine.
I think i could set spat.viewer without too much problem but i have some doubt about how would reac oper. There is no radial concept here right? What would you be your advice when working with this kind of set up? however the spat library is quite rich, i didn’t find any exemple of complex set up of speaker array that are not spherical. I see this octo exemple, but it’s quite simple too.

Also how would you manage speaker directivity in spat? I guess by default, the speaker is directed towards the virtual listener. Even if the speaker is place bellow the listener? Do we have a way to control that ? maybe it just doesn’t have influence in the calculation?

-eric

#2

Hi Eric,

Good hear you have access to a nice speaker setup. Enjoy!

In the spat paradigm, the speakers are supposed to be on a sphere (or circle) and oriented towards the sweet spot.

  • if the transducers are not on a sphere, but rather on a cube, you can use [spat.align~] to virtually re-align (in terms of delays and gains) the speakers so that they appear on a virtual sphere.
    For instance, with HOA, the workflow would be :
    [spat.pan~] (HOA encoder) --> [spat.decoder~] (HOA decoder, assuming a spherical setup) --> [spat.align~] (compensate for non-spherical layout) --> dac~
    (in the [spat.decoder~], only the azimuths and elevations of the speakers are taken into account, no matter the distances you provide in the Max messages)

Please note that this re-alignement is frequently done by the sound engineer, directly on the mixing desk. So, check with the sound engineer to make sure it is not done twice.

  • regarding speaker directivity, this is not taken into account. Spat assumes the speakers are somehow calibrated.

T.

#3

I understant the concept of speaker alignent using delays/gain - but here my issue is that it’s an uncommon array of speakers. Let say you have 15 columns of 3 speakers, placed regularly and finally it looks likes a grid cube, where speakers are at meeting point of planes. So we don’t have in spat a model capable of handle this kind of set up right? it’s not about correcting delays and gain here i thingk. it rather looks of a matrix custom set up . They actually have a pretty efficient way to manage it, but i really wanted to test efficiency of spat in there, with which i also got quite confortable lately.