“To clarify”: Techniques such as DirAC, Harpex etc. do upmixing,
usually via separating into direct and background parts. (HarpeX is
actually the exception.)
Mpeg compression of HOA (Mpeg 3DA) works in a similar way, although
there are certain differences.
I somewhat refuse to write a lot more, because it seems you would have
to study the field a bit more.
(After you might not write things such as “there is no upmixing in
ambisonics”.)
In my understanding, neither DirAC nor HarpeX have < anything > to do
with beamforming. ( At least not in some 1:1 way.)
Running away to play a concert, hopefully listened to at very high order...
Stefan Schreiber
----- Mensagem de Hannes Helmholz <hannes.helmh...@chalmers.se> ---------
Data: Fri, 23 Oct 2020 18:27:10 +0200
De: Hannes Helmholz <hannes.helmh...@chalmers.se>
Assunto: Re: [Sursound] ORTF-3D With Higher-order Ambisonics
Para: Surround Sound discussion group <sursound@music.vt.edu>, Stefan
Schreiber <st...@mail.telepac.pt>
To clarify. There is no "upmixing" in Ambisonics (at least none that
does not involve more sophisticated beamforming and therefore
changes the spatial properties of the captured sound field).
In general, the number of sensors of the array determines the
maximum Spherical Harmonic order ("B-format") that can be extracted
from the raw microphone signals ("A-format"). That is
(order+1)^2=mics for the case that the sensors are arranged
according to a Fliege-Meier grid, which is the most efficient
arrangement.
Higher Ambisonics orders could be extracted from the captured
signals. However, these "B-Format" signals will contain an
increasing amount of spatial aliasing (boost towards higher
frequencies). Such a higher-order decomposition will have seemingly
higher spatial resolution, however the additional information is
composed purely out of artifacts (the spatial aliasing) that have
nothing to do with the original sound field. Therefore, extracting
or rendering at a higher order than the microphone array allows is
mathematically possible but conceptually not beneficial and possibly
perceptually unfavorable.
In that way, the OctaMic seems to provide a reasonable solution for
second order nevertheless. To be pedantic, the frequency range in
which some apparent microphone signals can be __accurately__
transformed into the desired Ambisoncs (SH) representation depends
on many aspects. There not only the physical arrangement, but also
the employed processing plays an important role. The decomposition
is no hidden secret but still can be performed with different care
or aspects in mind (especially in terms of the employed modal radial
filters which compensate for the scattering of the microphone
array). Therefore it could be interesting to also experiment with
different software solutions there in order to yield different
results.
Nevertheless, there is one aspect to distinguish when looking solely
at the reproduction part of an Ambisonics pipeline. A first order
signal for example can be reproduced without drawbacks (and actually
without any additional processing) at second order. However, one
should not term this "upmixing", since the higher order components
("B-Format") of the Ambisoncs signals will simply be left at zero
(since this information is not available).
Such decoded reproduction signals could nevertheless utilize the
higher dentistry of for example a second order loudspeaker grid. If
this actually yields any technical or perceptual benefits I don't
know, since I am not so familiar with loudspeaker reproduction.
Regarding binaural rendering "directly in the SH domain", there is
no reason or benefit to render at a higher order than the source
material. For the case that the binaural renderer follows the
"virtual loudspeaker" approach, the only limiting factor of the
reproduction resolution (following the resolution of the source
material) would be the utilized HRTF data set.
In any case, "upmixing" or decomposition of the microphone signals
at a higher order than the array architecture supports is not
reasonable.
I also noticed that people are arguing over different ways on how to
utilize the microphone signals that you have available. To be frank,
lower order signals, especially first and even second order, are not
great in preserving the spatial properties of the captured sound
field. That is true in terms of instrumental accuracy due to
physical and mathematical limitations, but also perceptual fidelity.
The limitations have different implications for either reproduction
technique, e.g. spatial resolution, preservation of timbre, source
externalization, etc..
However, low order signals is what we mostly (or only begin with to)
have available at the moment. But since there is no universally
"right" way to treat them, the only option is to make things sound
as "nice" as possible at the end of the day. :)
I hope this was useful to some to understand the decomposition of
the microphone signals and the connected implications a bit better.
Hannes Helmholz
(PhD Student, somewhere)
Can make mistakes. Also just human. Please forgive.
On 2020-10-23 15:10, Stefan Schreiber wrote:
I think COMPASS should upsample from 2nd order to 3rd order (and higher).
However, the Octomic is not giving full 2nd order. (2h1v)
You still could set the missing element (FuMa R element, ACN 8) to
0 and do this.
Has anybody tried this?
Best,
Stefan
----- Mensagem de Steven Boardman <boardroomout...@gmail.com> ---------
Data: Fri, 23 Oct 2020 14:04:21 +0100
De: Steven Boardman <boardroomout...@gmail.com>
Assunto: Re: [Sursound] ORTF-3D With Higher-order Ambisonics
Para: Surround Sound discussion group <sursound@music.vt.edu>
Hi Len
I know Harpex is only first order, and a calibrated 2nd order mic
will produce better first order patterns. I was referring to just
Tetramic when I mentioned 1st order...
My point about Harpex was; you tried to do ORTF with only one
Octomic, and it wasn’t good.
Did you try to do 3D ORTF with 2 Octomic, experimenting with
positions, and which position each Harpexed Octomic synthesises?
Maybe Octomics on an opposing corner with Harpex from each
decoding alternately into the other corners?
The two Octomics could each have the 2 straightforward decodes (to
obtain 4 vertexes). But each of the other corners have 1 spaced
decode from each Octomic.
So the other vertexes are more deco-related from using alternate
Octomic spaced Harpex decodes.
Hope that makes sense, or shoot me down with bag of crap…:)
PS don’t suppose you have a super-cardioid response of the
Octamic? The CCM41 capsules used in the 3D ORTF are excellent,
with pretty dam good polar response upto 16k. (Obviously they
wouldn’t be anyway near the same in an array being so close to
each other:)
https://schoeps.de/en/products/ccm/ccm-microphones/supercardioids/ccm-41.html
On 23 Oct 2020, at 03:16, moskowitz <lenmoskow...@optonline.net> wrote:
Steven Boardman wrote:
Have you tried it with first order?
I would of thought using one Harpexed octomic to synthesise all
the spaced
mics, would be a bit blurry, did you try with two?
Harpex only operates at first-order.
We used OctoMic's first-order B-format as input to Harpex.
We've used two OctoMics for ORTF. Since OctoMic's cardioid
patterns are so stable and consistent, and because pointing
angles can be fine-tuned in post, two OctoMics make probably the
world's finest ORTF array.
Len Moskowitz (mosko...@panix.com)
Core Sound LLC
www.core-sound.com
Home of OctoMic and TetraMic
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe
here, edit account or options, view archives and so on.
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe
here, edit account or options, view archives and so on.
----- Fim da mensagem de Steven Boardman <boardroomout...@gmail.com> -----
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe
here, edit account or options, view archives and so on.
----- Fim da mensagem de Hannes Helmholz <hannes.helmh...@chalmers.se> -----
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.