17 Dec
2013
17 Dec
'13
5:13 a.m.
All,
Some preliminary research reveals several FOSS implementations of WFS
(wave field synthesis). What is not entirely clear is how these
implementations stack up to something like Sonic Emotion. I presume they
will be subpar but the question is by how much and in what ways?
Ok, now another question. Is anyone aware of a 3D FOSS WFS
implementation (multiple horizontal rows) and how hard would it be to use?
Best wishes,
--
Ivica Ico Bukvic, D.M.A
Composition, Music Technology
Director, DISIS Interactive Sound & Intermedia Studio
Director, L2Ork Linux Laptop Orchestra
Head, ICAT IMPACT Studio
Virginia Tech
Department of Music
Blacksburg, VA 24061-0240
(540) 231-6139
(540) 231-5034 (fax)
disis.music.vt.edu
l2ork.music.vt.edu
ico.bukvic.net
17 Dec
17 Dec
11:29 a.m.
On Mon, Dec 16, 2013 at 11:13:18PM -0500, Ivica Ico Bukvic wrote:
Some preliminary research reveals several FOSS
implementations of
WFS (wave field synthesis). What is not entirely clear is how these
implementations stack up to something like Sonic Emotion. I presume
they will be subpar but the question is by how much and in what
ways?
Ok, now another question. Is anyone aware of a 3D FOSS WFS
implementation (multiple horizontal rows) and how hard would it be
to use?
Given the number of speakers used, the S.E. system can't be pure
WFS except at very low frequencies. It probably uses a combination
of techniques: WFS, some things based on AMB theory, delays, etc.
but they won't tell you more.
But the main difference to open source systems are to be found not
in the rendering system, but in the one used to create and define
the content. In commercial systems this will be very visual, hide
the technicalities, and probably integrate with protools. It will
also be closed and let you do predefined things only. All that
makes it easier to use for the non-expert.
Open source systems tend to provide less in this area, but will
have interfaces that allow you to define your own production
workflow and tools, usually via OSC. For example, the system
which I developed and installed in Parma will let you control
the position and smooth movements of virtual sources via OSC,
but little more. Anything else has to be build on top of this.
The main tool used here in Parma is a 'mixer' that instead of
really mixing its inputs, controls the rendering engine instead,
while also taking care of changing e.g. reverb levels and delays
in function of source position. For static sources that is
almost everything you need, apart from standard production
tools. For more dynamic setups I either write ad-hoc code
(usually Python, but you could use SC, Pd, Csound...), or
plugins sending OSC from automation tracks in Ardour.
A WFS system using two or more rows wouldn't be real 3D WFS,
the vertical component would use conventional panning. A real
3D WFS system would require filling the walls with speakers.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
4:14 p.m.
On Dec 17, 2013 5:29 AM, "Fons Adriaensen" <fons(a)linuxaudio.org> wrote:
On Mon, Dec 16, 2013 at 11:13:18PM -0500, Ivica Ico Bukvic wrote:
I had an opportunity to listen to their system and even though it was quite
sparse it still delivered a very compelling image even close to the
speakers. If the system was using anything in the way of ambisonics it
would have done so only using the actual horizontal array. Given that
highest repreducible frequency is directly related to the distance between
the speakers I am wondering if this may be because they are also trying to
render waves from virtual speakers as they propagate through the real
speakers as well as using a selection of speakers to render certain sounds
as per recent publications in this area.
Some preliminary research reveals several FOSS
implementations of
WFS (wave field synthesis). What is not entirely clear is how these
implementations stack up to something like Sonic Emotion. I presume
they will be subpar but the question is by how much and in what
ways?
Ok, now another question. Is anyone aware of a 3D FOSS WFS
implementation (multiple horizontal rows) and how hard would it be
to use?
Given the number of speakers used, the S.E. system can't be pure
WFS except at very low frequencies. It probably uses a combination
of techniques: WFS, some things based on AMB theory, delays, etc.
but they won't tell you more.
But the main difference to open source systems are to be found not
in the rendering system, but in the one used to create and define
the content. In commercial systems this will be very visual, hide
the technicalities, and probably integrate with protools. It will
also be closed and let you do predefined things only. All that
makes it easier to use for the non-expert.
Open source systems tend to provide less in this area, but will
have interfaces that allow you to define your own production
workflow and tools, usually via OSC. For example, the system
which I developed and installed in Parma will let you control
the position and smooth movements of virtual sources via OSC,
but little more. Anything else has to be build on top of this.
The main tool used here in Parma is a 'mixer' that instead of
really mixing its inputs, controls the rendering engine instead,
while also taking care of changing e.g. reverb levels and delays
in function of source position. For static sources that is
almost everything you need, apart from standard production
tools. For more dynamic setups I either write ad-hoc code
(usually Python, but you could use SC, Pd, Csound...), or
plugins sending OSC from automation tracks in Ardour.
A WFS system using two or more rows wouldn't be real 3D WFS,
the vertical component would use conventional panning. A real
3D WFS system would require filling the walls with speakers.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
http://lists.linuxaudio.org/listinfo/linux-audio-user
5:04 p.m.
On Tue, Dec 17, 2013 at 10:14:45AM -0500, Ivica Bukvic wrote:
I had an opportunity to listen to their system and
even though it was quite
sparse it still delivered a very compelling image even close to the
speakers. If the system was using anything in the way of ambisonics it
would have done so only using the actual horizontal array. Given that
highest repreducible frequency is directly related to the distance between
the speakers I am wondering if this may be because they are also trying to
render waves from virtual speakers as they propagate through the real
speakers as well as using a selection of speakers to render certain sounds
as per recent publications in this area.
'Filling the gaps' with virtual speakers doesn't change things - for the
same reason that you can't pull yourself up by your bootstraps.
Consider the simple case of a plane wave from a direction orthogonal
to the line of speakers. All speakers will get the same signal. To
create a virtual speaker between every two adjacent real ones, those
two would just have to get more of the same signal. So nothing changes,
except at the ends of the line (and you may rediscover tapering).
Ambisonic reproduction will create images 'projected on the line of
speakers' a virtual source in between the speakers will appear at
the same distance as the real speakers. It's possible to change this
using near-field compensation, but only for low frequencies.
Distance illusions can be created by exploiting psycho-acoustic
effects rather than physics. I suspect systems such as S.E. are
particularly good at this.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
11:14 p.m.
-----Original Message-----
From: Fons Adriaensen [mailto:fons@linuxaudio.org]
Sent: Tuesday, December 17, 2013 11:04 AM
To: Ivica Bukvic
Cc: A list for linux audio users
Subject: Re: [LAU] question about FOSS WFS implentations
On Tue, Dec 17, 2013 at 10:14:45AM -0500, Ivica Bukvic wrote:
> I had an opportunity to listen to their system and even though it was
quite
> sparse it still delivered a very compelling image
even close to the
> speakers. If the system was using anything in the way of ambisonics it
> would have done so only using the actual horizontal array. Given that
> highest repreducible frequency is directly related to the distance
between
> the speakers I am wondering if this may be
because they are also trying
to
> render waves from virtual speakers as they
propagate through the real
> speakers as well as using a selection of speakers to render certain
sounds
as per recent
publications in this area.
'Filling the gaps' with virtual speakers doesn't change things - for the
same reason that you can't pull yourself up by your bootstraps.
Consider the simple case of a plane wave from a direction orthogonal
to the line of speakers. All speakers will get the same signal. To
create a virtual speaker between every two adjacent real ones, those
two would just have to get more of the same signal. So nothing changes,
except at the ends of the line (and you may rediscover tapering).
Ambisonic reproduction will create images 'projected on the line of
speakers' a virtual source in between the speakers will appear at
the same distance as the real speakers. It's possible to change this
using near-field compensation, but only for low frequencies.
Distance illusions can be created by exploiting psycho-acoustic
effects rather than physics. I suspect systems such as S.E. are
particularly good at this.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
18 Dec
18 Dec
1:28 a.m.
On Tue, Dec 17, 2013 at 05:14:08PM -0500, Ivica Ico Bukvic wrote:
What if the denser line is positioned a distance
behind the original
speakers (as in the entire space is actually bigger with a denser array
encircling the real physical array)? Sure, this would do nothing for an
orthogonal direction, but it may improve the perception of other angles, no?
And how would that distant denser array (of virtual speakers) be
created ? By WFS ? That assumes that the WFS system actually works
(as WFS) at the frequency of interest. But the starting point is
that it doesn't. This is again trying to pull yourself up by your
bootstraps.
I'm pretty sure that S.E. does some clever things above the aliasing
frequency. I'm also pretty sure they can produce convincing distance
cues (for some material, probably carefully chosen for the demos).
But I don't believe that any of that is actually WFS.
There were some experiments (in the early 1950s IIRC) using just
three speakers on stage, playing a recording of an orchestra
*made on the same stage*. I'm too young to have been there, but
from what I've read it seems the reproduction was near to perfect.
That doesn't mean that three speakers on stage can do reproduce
whatever you want with the same result.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
17 Dec
17 Dec
6:15 p.m.
On 12/17/2013 05:13 AM, Ivica Ico Bukvic wrote:
All,
Some preliminary research reveals several FOSS implementations of WFS
(wave field synthesis). What is not entirely clear is how these
implementations stack up to something like Sonic Emotion. I presume they
will be subpar but the question is by how much and in what ways?
Ok, now another question. Is anyone aware of a 3D FOSS WFS
implementation (multiple horizontal rows) and how hard would it be to use?
Best wishes,
Hi,
i heard some of the FOSS implementations had Problems with moving
sources, creating really noticeable artifacts, have you tried some
current ones? Is that still the case?
Greetings
Markus
3790
days inactive
3791
days old
linux-audio-user@lists.linuxaudio.org
6 comments
4 participants
participants (4)
-
Fons Adriaensen -
Ivica Bukvic -
Ivica Ico Bukvic
-
Markus Seeber