Group Admins

  • Profile picture of warusfel
  • Profile picture of T.

Spat

Public Group active 1 week, 1 day ago

User group for Spat, panoramix, ToscA , and ADMix

Spat.decoder bformat channel weighting

Author 2 Subscribed Users |
Profile photo of natashabarrett
natashabarrett

Hi. In spat.decoder @type=bformat, what is the channel weighting / normalisation? Is it FuMa or SN3D? I don’t seem to find this clarified in the documentation. Or in other words, if you use the spat.decoder for a FuMa source (or a Soundfield recording) do you need to divide W by 0.707107? This came up while using bformat2UHJ setting, where I am now unsure of what channel weighting the bformat2UHJ decoder is expecting. Unlike spat pan, the decoder can’t be set with an @norm argument. Thanks for info, apologies if this question has come up and I’ve missed the answer.

September 6, 2016 at 09:38 #18953
Profile photo of T.
T.

Hi Natasha,

This normalization is described in the spat.harmonics patcher.
You’ll see that “@type = bformat” corresponds neither to FuMa nor SN3D. It is provided only for backward compatibility with legacy versions of Spat~.
I would recommend not to use it.

As for the impact of normalization on UHJ transcoder, that’s a good question.
I’ll have to think about it.

T.

September 6, 2016 at 15:29 #18964
Profile photo of natashabarrett
natashabarrett

I see. So to be clear then, the bformat2UHJ transcoder is expecting spat’s own flavour of bformat, where all four channels have a different normalisation to anything else? (i.e. needing a manual conversion as spat-flavour bformat is not an option in hoaconverter). Or is bformat2UHJ expecting FuMa?

Thanks
Natasha

September 6, 2016 at 16:52 #18969
Profile photo of T.
T.

I’d say with 95% confidence that the UHJ transcoder expects FuMa.
For the remaining 5%, I need to clear my head and think about it…

T.

September 6, 2016 at 17:20 #18970
Profile photo of T.
T.

Also, two things to be careful about :
– if you inspect the inlets of [spat.decoder~ @numinputs 4 @numoutputs 2 @type bformat2uhj], you’ll see that it expects W,X,Y,Z signals. So, this does not follow ACN channel ordering.
– the bformat2uhj follows the spat coordinate convention where X axis points toward the right, and Y axis points forward. In most other HOA tools, X is pointing forward and Y to the left. So, you might need a 90° rotation…

— Pasted Max Patch, click to expand. —
September 6, 2016 at 23:27 #18978
Profile photo of natashabarrett
natashabarrett

Hi Thibaut, Thanks for the patch.

I have a question refinement: prior to the recent new release of the bformat2uhj transcoder (in April?) you mentioned that when specifying inputs=4 and outputs=2, the transcoder would include energy from the vertical component. But when I compare inputs=4 and inputs=3, then elevate the signal, the outputs are identical. I was wondering whether you had removed the vertical energy feature, or if there is something else going on? (e.g. adding vertical energy equally over left and right output channels then compensating to avoid an overall gain increase?)

Thanks
Natasha

September 7, 2016 at 08:33 #18979
Profile photo of T.
T.

April ? I have no recollection of that…
Indeed the Z component is not taken into account in the L/R mix.
That would be a nice feature. If you have any publication/equation, I would be glad to implement it.

T.

September 7, 2016 at 09:30 #18980

You must be logged in to reply to this topic.

Log in now