When to set the resolution

[Richard Lee]

While I'm aware of the problem that our GV has brought up, I don't think this
is within the scope of a recording format.

The fundamental principal of Ambisonics is that the recording/storage format is
a DEFINITION of what the sound field should be.

How you reproduce it is a separate problem. ie the Decoder problem. The
decoder should take into account psychoacoustics, the size of the Mk1HH, number
of speakers, room, the phase of the moon etc. But the DEFINITION should NOT be
concerned with such matters.

There are a number of things which I want to try out if and when I ever get a
modern reliable machine/compiler capable of compiling VideoLAN. These
including optimising for 1st order and using the higher orders rather more
sparingly than if full soundfield recreation AT A POINT is required. On first
glance, this would seem to be congruent with Large Area High Order decoders.

Of course this is all pontificating at present so someone needs to try it out.

But we should keep the recording format as simple and clean as possible with
the minimum of metadata.

[e deleflie]

Richard,

> The fundamental principal of Ambisonics is that the recording/storage format is
> a DEFINITION of what the sound field should be.
>
> How you reproduce it is a separate problem.

But we have introduced a new complexity ... which is how to mix 1st
order with higher orders. I'm wondering how that fits into the
fundamental principles.

Perhaps the real fix is to simply not attempt to mix higher orders
with 1st order recorded material. But I dont think this is feasible.
People will do it anyway. It would be better to work out some way of
catering to it. What are your thoughts on this?

The current discussion is hinting at a format that has 3 different
sets of channel counts:

1) 1st order (4channels)
2) Higher Order (16 channels ... 3rd order periphonic)
3) Mixed orders (20 channels .... first 16 are HOA, last 4 are 1st order).

Messy... but simpler than having a dozen different possible channel
counts ... all catering for different mixed orders. It needs to be
simpler still.

If there was a way to have an ambisonics format that _always_ had 16
channels... and could mix first order with higher order.... I think we
would be home and hosed.

Etienne