31 Jul
2010
31 Jul
'10
4:40 a.m.
Folks,
I'm a recovering audiophile. When i was reading the magazines and
reading about over priced (in my now opinion) speakers, the words "full
range" tended to mean that a speaker was reasonably flat from 20 Hz to
20 kHz. Granted those were unusual.
I have read that speakers in an ambisonic set up should be "full range".
I'd like to set up a ambisonic speaker system (8 channel to start), and
the prospect of 8 full range channels is daunting. Since it seems they
would be stand or wall mounted (at least some of them) that means
monitors and subwoofers. Since all channels must be the same, that means
8 subwoofers...somewhere in the room.
So what does "full range" mean usually and what does it mean in terms of
talk on this list?
Thanks,
Bearcat
31 Jul
31 Jul
11:04 a.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 07/31/2010 04:40 AM, "Bearcat M. Şandor" wrote:
Folks,
I'm a recovering audiophile.
:-]
I have read that speakers in an ambisonic set up
should be "full range".
I'd like to set up a ambisonic speaker system (8 channel to start), and
the prospect of 8 full range channels is daunting. Since it seems they
would be stand or wall mounted (at least some of them) that means
monitors and subwoofers. Since all channels must be the same, that means
8 subwoofers...somewhere in the room.
So what does "full range" mean usually and what does it mean in terms of
talk on this list?
first, as initial shock therapy to the (ex-?) audiophile: ambisonics is
not full-range in the usual sense. its localisation mechanism only works
in a very limited frequency range. the idea is to use that small range
to produce such good cues to the brain that it conveniently ignores that
the rest is garbage ;)
i'm taking localisation only - timbre reproduction must of course happen
from 20 to 20k...
if you want eight speakers, that would fix you up for up to third order.
since i assume you're not much into computer-generated music (which
would provide some source material), and since i know that higher-than
1st-order recordings are rare (because i've personally made at least 1%
of them...), it could make sense to go for 6 speakers initially. very
good for first-order replay, very good for "super-stereo", easily
accommodates 5.1 content with only minor placement errors, easier to
place, and 25% cheaper. unless you choose a very exotic brand and model,
it should be no problem to extend this setup to eight or more later.
it's no problem (and in fact customary) to use subwoofers with
ambisonics for low-range extension. you can either forego localisation
in that band and just use a mono sub (perfectly ok), use 2 and have some
l/r signal in ithem, or you can use 4 subs in the corners of the room
and drive them with a special horizontal decode. that's pretty awesome,
because you can do a decode such that one speaker pushes while the
opposite speaker pulls. of course that wastes quite some amp power, but
it makes your room effectively infinite volume, which means you can fit
very long waves into it without weird effects, and you will win a nice
trouser-flapping effect without boominess and that obnoxious pressure on
the ears.
if you're going to use ambdec for decoding, i can show you how to create
two separate configurations for the full-range speakers and the subs,
and to combine them into a single matrix (that's assuming that the subs
themselves do their own low-pass filtering - if not, you'd need to do
that with a plugin, but that means you will have to use another
application in addition to ambdec).
best,
jörn
1:26 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Sat, Jul 31, 2010 at 11:03:03AM +0200, Jörn Nettingsmeier wrote:
On 07/31/2010 04:40 AM, "Bearcat M. Şandor"
wrote:
This is very misleading. Field reconstruction is limited to the
LF range, but that does not mean that Ambisonics has no localisation
paradigm for the rest of the frequency range. It has, and it consists
of creating an energy vector in the right direction and the best
possible magnitude. This is the same as used by intensity based
stereo and VBAP. Compared to the latter Ambisonics does a better
job as it will avoid giving preferential treatment to the speaker
positions.
Ciao,
--
FA
There are three of them, and Alleline.
So what does "full range" mean usually
and what does it mean in terms of
talk on this list?
first, as initial shock therapy to the (ex-?) audiophile: ambisonics
is not full-range in the usual sense. its localisation mechanism
only works in a very limited frequency range. the idea is to use
that small range to produce such good cues to the brain that it
conveniently ignores that the rest is garbage ;)
11:14 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 07/31/2010 03:03 AM, Jörn Nettingsmeier wrote:
first, as initial shock therapy to the (ex-?)
audiophile: ambisonics is
not full-range in the usual sense. its localisation mechanism only works
in a very limited frequency range. the idea is to use that small range
to produce such good cues to the brain that it conveniently ignores that
the rest is garbage ;)
i'm taking localisation only - timbre reproduction must of course happen
from 20 to 20k...
if you want eight speakers, that would fix you up for up to third order.
since i assume you're not much into computer-generated music (which
would provide some source material), and since i know that higher-than
1st-order recordings are rare (because i've personally made at least 1%
of them...), it could make sense to go for 6 speakers initially. very
good for first-order replay, very good for "super-stereo", easily
accommodates 5.1 content with only minor placement errors, easier to
place, and 25% cheaper. unless you choose a very exotic brand and model,
it should be no problem to extend this setup to eight or more later.
it's no problem (and in fact customary) to use subwoofers with
ambisonics for low-range extension. you can either forego localisation
in that band and just use a mono sub (perfectly ok), use 2 and have some
l/r signal in ithem, or you can use 4 subs in the corners of the room
and drive them with a special horizontal decode. that's pretty awesome,
because you can do a decode such that one speaker pushes while the
opposite speaker pulls. of course that wastes quite some amp power, but
it makes your room effectively infinite volume, which means you can fit
very long waves into it without weird effects, and you will win a nice
trouser-flapping effect without boominess and that obnoxious pressure on
the ears.
if you're going to use ambdec for decoding, i can show you how to create
two separate configurations for the full-range speakers and the subs,
and to combine them into a single matrix (that's assuming that the subs
themselves do their own low-pass filtering - if not, you'd need to do
that with a plugin, but that means you will have to use another
application in addition to ambdec).
best,
jörn
Jörn,
Thank you. I think i understand all that, but let me take this apart to
make sure. What you're saying is that having full range speakers only
effects the playback quality of the music not the ambisonics and
that ambisonics itself does not *need* full range speakers, but that
having full range speakers is better than not. Is that correct?
I think part of my confusion is that i'm still thinking of it as having
6 or 8 or more *channels* when that is not the case. The speakers are
not steered, they are driven. So you are not going to have a situation
where the speakers behind you are only reproducing high to midrange
information as was the case with Dolby pro-logic or something. Right?
I am planning on subs that have their own low-pass filters. I have a
pair of Anthony Gallo Reference 3.1s but for this ambisonics set up, i'd
get 4 pairs of the Anthony Gallo Stradas and 4 T3 subs. stradas:
http://www.roundsound.com/reference-strada.htm t3:
http://www.roundsound.com/tr-3-subwoofers.htm Most of the stuff on
those pages is marketing of course, but the satellites have a range of
45 Hz - 20 Khz +- 3 db and the subs go down to 22 Hz (no variance given).
That push-pull idea sounds fantastic. I would love it if you'd show me
how to create a set up for a system like that using ambdec, if you have
the inclination and the time. That's what i was planning on using anyhow.
Thank you very much Jörn
Bearcat
1 Aug
1 Aug
1:39 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 07/31/2010 11:14 PM, "Bearcat M. Şandor" wrote:
Thank you. I think i understand all that, but let me
take this apart to
make sure. What you're saying is that having full range speakers only
effects the playback quality of the music not the ambisonics and
that ambisonics itself does not *need* full range speakers, but that
having full range speakers is better than not. Is that correct?
ah, no, sorry. as fons remarked earlier, that mail was kind of sloppy.
it's just that when you embark on the ambisonic trip, there are many
high-end stereo truisms you should be prepared to let go of.
worse yet, ambisonic systems have some clear disadvantages that would
not be accepted in high-end stereo circles - different localisation
ambiguities, sometimes coloration, sometimes extreme phasing, etc.
and there is no reason to accept them in stereo.
at the same time, there is no reason to reject a system that displays
them, because it's the price of good surround sound. you don't reject
stereo for its lack of rear localisation, either...
I think part of my confusion is that i'm still
thinking of it as having
6 or 8 or more *channels* when that is not the case. The speakers are
not steered, they are driven. So you are not going to have a situation
where the speakers behind you are only reproducing high to midrange
information as was the case with Dolby pro-logic or something. Right?
no, never.
I am planning on subs that have their own low-pass
filters. I have a
pair of Anthony Gallo Reference 3.1s but for this ambisonics set up, i'd
get 4 pairs of the Anthony Gallo Stradas and 4 T3 subs. stradas:
http://www.roundsound.com/reference-strada.htm t3:
http://www.roundsound.com/tr-3-subwoofers.htm Most of the stuff on
those pages is marketing of course, but the satellites have a range of
45 Hz - 20 Khz +- 3 db and the subs go down to 22 Hz (no variance given).
That push-pull idea sounds fantastic. I would love it if you'd show me
how to create a set up for a system like that using ambdec, if you have
the inclination and the time. That's what i was planning on using anyhow.
basically, you take fons' example for the octagon, set the speakers up
at the correct angles and then enter the actual distances into the
matrix, so that delay and near-field compensation is correct for your
setup. then you take the example square for the subs, set them into the
corners of the room and again enter the correct distances.
at this point you have two ambdec instances running, one for the tops,
one for the subs. you'd need to adjust the relative loudness with
ambdec's faders. once you've determined the correct relative levels, you
can factor them into the matrix coefficients of the sub decoder and run
both ambdecs at 0dB. now you can hack both matrices into one and end up
with a 12 channel ambdec configuration: 8 tops and four subs. easier to
start up in daily use.
3 Aug
3 Aug
7:46 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Sun, 2010-08-01 at 13:39 +0200, Jörn Nettingsmeier wrote:
On 07/31/2010 11:14 PM, "Bearcat M. Şandor"
wrote:
Hmmm, so, you send WXYZ to the four subs and whatever order you can
decode to the regular speakers (6, 8, etc). Sounds good.
What if you want to include a crossover? (that's exactly what I'm
working on right now in the openmixer software).
So, you high pass WXYZ and send that to the regular speakers, low pass
WXYZ and send that to the subs... and what do you do with the rest of
the Ambisonics components? I imagine I would just send the full
frequency range to the regular speakers and let them do the best they
can, right?
-- Fernando
I am planning on subs that have their own
low-pass filters. I have a
pair of Anthony Gallo Reference 3.1s but for this ambisonics set up, i'd
get 4 pairs of the Anthony Gallo Stradas and 4 T3 subs. stradas:
http://www.roundsound.com/reference-strada.htm t3:
http://www.roundsound.com/tr-3-subwoofers.htm Most of the stuff on
those pages is marketing of course, but the satellites have a range of
45 Hz - 20 Khz +- 3 db and the subs go down to 22 Hz (no variance given).
That push-pull idea sounds fantastic. I would love it if you'd show me
how to create a set up for a system like that using ambdec, if you have
the inclination and the time. That's what i was planning on using anyhow.
basically, you take fons' example for the octagon, set the speakers up
at the correct angles and then enter the actual distances into the
matrix, so that delay and near-field compensation is correct for your
setup. then you take the example square for the subs, set them into the
corners of the room and again enter the correct distances.
at this point you have two ambdec instances running, one for the tops,
one for the subs. you'd need to adjust the relative loudness with
ambdec's faders. once you've determined the correct relative levels, you
can factor them into the matrix coefficients of the sub decoder and run
both ambdecs at 0dB. now you can hack both matrices into one and end up
with a 12 channel ambdec configuration: 8 tops and four subs. easier to
start up in daily use.
8:17 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 08/03/2010 07:45 PM, Fernando Lopez-Lezcano wrote:
Hmmm, so, you send WXYZ to the four subs and whatever order you can
decode to the regular speakers (6, 8, etc). Sounds good.
it does :) except you'd usually throw away z, unless you wanted to try a
tetrahedral setup for the subs (for example: front left floor, front
right ceiling, rear left ceiling, rear right floor). this setup is known
to be unstable for full-range use, but could work great for bass,
depending on how far your subs have to reach up. then again, i'm not
sure if the rigging hassle is worth the z information in the bass.
What if you want to include a crossover? (that's
exactly what I'm
working on right now in the openmixer software).
So, you high pass WXYZ and send that to the regular speakers, low pass
WXYZ and send that to the subs... and what do you do with the rest of
the Ambisonics components? I imagine I would just send the full
frequency range to the regular speakers and let them do the best they
can, right?
hmm, i think i would rely on the internal electronics of the tops for
the HPF and feed them a full-range signal of all components, and then
maybe measure one and design a suitable roll-off for the bass
speakers... but i have never had the luxury to do that, i always set the
LPF by ear.
sending high-passed WXYZ and full-range higher orders to the tops won't
work well, i guess. if you want to take care of the band-splitting in
your mixer, your best bet would be ganged high/low-pass filters for all
the orders. if you have the time, gear, and expendable ph.d. student
labor at hand, it would be fun to try three- or four-way systems with an
increasing number of speakers as the frequency goes up.
but since the most expensive part for massive ambi systems is the
amping, that's probably an economic dead-end, at least for now.
best,
jörn
9:38 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Tue, Aug 03, 2010 at 08:17:03PM +0200, Jörn Nettingsmeier wrote:
sending high-passed WXYZ and full-range higher orders
to the tops
won't work well, i guess. if you want to take care of the
band-splitting in your mixer, your best bet would be ganged
high/low-pass filters for all the orders. if you have the time,
gear, and expendable ph.d. student labor at hand, it would be fun to
try three- or four-way systems with an increasing number of speakers
as the frequency goes up.
but since the most expensive part for massive ambi systems is the
amping, that's probably an economic dead-end, at least for now.
A separte VLF decoder is one of the features I want in the next
generation Ambdec. The question I'm asking myself is how much
'configurability' this requires. It's no problem adding the
filters that would perform a 'perfect' split of VLF and the
rest. But then you have to consider the response of the units
themselves - many subs have their own filters, and the LF
response of the 'full range' speakers can't be assumed to be
really flat in the crossover region. A really good xover has
to take into account the response of the two systems it is
driving. All suggestions are welcome...
Another question to two (Joern and Fernando) AMB nerds: does
a multicnannel AMB aware dynamics processor make sense ? When
would you use it ? I'm asking because one of the plugins I'm
developing could be exactly that. ATM it's just compression
(with real RMS response), but the full version will have a
noise gate as well, Do you need such a thing ? Current plans
are that the full version will be mono/stereo only.
Ciao,
--
FA
There are three of them, and Alleline.
11:11 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 08/03/2010 09:38 PM, fons(a)kokkinizita.net wrote:
A separte VLF decoder is one of the features I want in
the next
generation Ambdec. The question I'm asking myself is how much
'configurability' this requires.
as far as i'm concerned, it would be cool to have some help creating a
"hybrid" matrix like i described to bearcat, to avoid having to run
separate ambdecs for tops and subs. the most important thing would be
another gain coefficient to match the subs to the tops, similar to what
you already have to tweak the two psychoacoustic bands.
It's no problem adding the
filters that would perform a 'perfect' split of VLF and the
rest. But then you have to consider the response of the units
themselves - many subs have their own filters, and the LF
response of the 'full range' speakers can't be assumed to be
really flat in the crossover region. A really good xover has
to take into account the response of the two systems it is
driving. All suggestions are welcome...
i don't think ambdec can do much about it, for the reasons you mention,
and for the fact that the subs will be physically misaligned wrt the
tops. so what we need is a delay (already there), and maybe, just maybe
some "+/-" buttons for the delay to help with a setup by ear, and a tone
generator that creates test signals in the critical overlapping band -
these could then simply be tuned for maximum loudness.
Another question to two (Joern and Fernando) AMB
nerds: does
a multicnannel AMB aware dynamics processor make sense ? When
would you use it ? I'm asking because one of the plugins I'm
developing could be exactly that.
oh, certainly. it would be divine to *not* be able to do evil sum
compression in ambisonics, ever, but sometimes, earthly matters require
it...
but my favourite solution to this approach would be for ardour to do the
right thing wrt side chaining (or at least allowing the chains to be
made manually). i'm told a3 can do it, and i'll be checking it out.
but yours would certainly a welcome addition to my tool chest.
ATM it's just compression
(with real RMS response), but the full version will have a
noise gate as well, Do you need such a thing ? Current plans
are that the full version will be mono/stereo only.
N-way side chaining would be cool, with just a single set of controls.
the interesting question is how the measuring should be done and the
orders interpreted. and here we go again, at normalisation and channel
order ;)
11:42 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Tue, Aug 03, 2010 at 11:11:42PM +0200, Jörn Nettingsmeier wrote:
Well I don't think a plugin host can ever do the right thing unless
it is AMB aware itself. For a multichannel compressor the gain has to
be the same in all channels, AMB or not. So you can't just replicate a
mono plugin. For the compressor I'm developing, in AMB mode only the W
channel affects the gain. A host can't arrange that, unless 1) the
compressor is multichannel by design and 2) it has a separate detector
(sidechain) input and 3) the host allows you to send just one channel
of a multichannel track to that special input. Maybe A3 can do that.
But it would just cover this simple case. It's much easier to provide
the function *inside* the plugin.
If you want to test it it's available (along with an EQ) if you accept
the NDA.
Ciao,
--
FA
There are three of them, and Alleline.
Another
question to two (Joern and Fernando) AMB nerds: does
a multicnannel AMB aware dynamics processor make sense ? When
would you use it ? I'm asking because one of the plugins I'm
developing could be exactly that.
oh, certainly. it would be divine to *not* be able to do evil sum
compression in ambisonics, ever, but sometimes, earthly matters
require it...
but my favourite solution to this approach would be for ardour to do
the right thing wrt side chaining (or at least allowing the chains
to be made manually). i'm told a3 can do it, and i'll be checking it
out.
4 Aug
4 Aug
8:12 a.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Tue, 2010-08-03 at 21:38 +0200, fons(a)kokkinizita.net wrote:
On Tue, Aug 03, 2010 at 08:17:03PM +0200, Jörn
Nettingsmeier wrote:
That would be great... right now I'm using a HPF/LPF pair for my tests
(butterworth 12dB/oct, supercollider) and placing them high enough that,
well, maybe both the loudspeakers and the subs are in a region that is
not right where they are transitioning (~70Hz).
But I'm not really sure how I can actually measure the bloody thing...
sending high-passed WXYZ and full-range higher
orders to the tops
won't work well, i guess. if you want to take care of the
band-splitting in your mixer, your best bet would be ganged
high/low-pass filters for all the orders. if you have the time,
gear, and expendable ph.d. student labor at hand, it would be fun to
try three- or four-way systems with an increasing number of speakers
as the frequency goes up.
but since the most expensive part for massive ambi systems is the
amping, that's probably an economic dead-end, at least for now.
A separte VLF decoder is one of the features I want in the next
generation Ambdec. The question I'm asking myself is how much
'configurability' this requires. It's no problem adding the
filters that would perform a 'perfect' split of VLF and the
rest. But then you have to consider the response of the units
themselves - many subs have their own filters, and the LF
response of the 'full range' speakers can't be assumed to be
really flat in the crossover region. A really good xover has
to take into account the response of the two systems it is
driving. All suggestions are welcome...
No easy answer. It looks like a complicated affair if you want to solve
it "perfectly". I would start with a "perfect" filter (assuming nice
loudspeakers) and leave further tweaking for other programs.
Another question to two (Joern and Fernando) AMB
nerds: does
a multicnannel AMB aware dynamics processor make sense ? When
would you use it ? I'm asking because one of the plugins I'm
developing could be exactly that. ATM it's just compression
(with real RMS response), but the full version will have a
noise gate as well, Do you need such a thing ? Current plans
are that the full version will be mono/stereo only.
This one I never thought about, but I guess it would become
indispensable once your native capture format is ambisonics. It would be
nice to have it but I don't have an application screaming for it right
away.
Unrelated question, does ambdec do third order vertical? Will it?
(I'm going to have zenith and nadir speakers in the Listening Room so it
would be able to do 3rd order h and v).
-- Fernando
12:21 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 08/04/2010 08:11 AM, Fernando Lopez-Lezcano wrote:
Unrelated question, does ambdec do third order
vertical? Will it?
(I'm going to have zenith and nadir speakers in the Listening Room so it
would be able to do 3rd order h and v).
me too! wouldn't even need a gui to create them, but it would be great
to be able to have ambdec process arbitrary matrices, maybe even up to
fifth order, so that it can be used for all those spiffy installations
we have seen lately (ircam, iem, etc...)
best,
jörn
3:01 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Wed, Aug 04, 2010 at 12:20:56PM +0200, Jörn Nettingsmeier wrote:
On 08/04/2010 08:11 AM, Fernando Lopez-Lezcano wrote:
That is definitely the direction things are taking - to remove any
artificial limits. Also the whole processing system (C++) will have
a Python API. So you can create custom decoders in Python. This will
also make it easy to add OSC or MIDI control for example.
I'm still in doubt what to do with the configuration files. Since adding
OSC is easy, they could just be the textual form of OSC they are now.
The other option is to skip this and turn the config files into active
Python code, directly using the API.
Integrating a Python interpreter with real-time code and event-driven
GUI code requires some trickery which I'll explore and hopefully get
working while in holiday (end of this month).
Ciao,
--
FA
There are three of them, and Alleline.
Unrelated question, does ambdec do third order
vertical? Will it?
(I'm going to have zenith and nadir speakers in the Listening Room so it
would be able to do 3rd order h and v).
me too! wouldn't even need a gui to create them, but it would be
great to be able to have ambdec process arbitrary matrices, maybe
even up to fifth order, so that it can be used for all those spiffy
installations we have seen lately (ircam, iem, etc...) 
4:56 p.m.
On Wed, Aug 4, 2010 at 8:01 AM, <fons(a)kokkinizita.net> wrote:
I'm still in doubt what to do with the
configuration files. Since adding
OSC is easy, they could just be the textual form of OSC they are now.
The other option is to skip this and turn the config files into active
Python code, directly using the API.
Integrating a Python interpreter with real-time code and event-driven
GUI code requires some trickery which I'll explore and hopefully get
working while in holiday (end of this month).
That seems like a good approach. There's a lot to be said for using a
complete programming language for configuration. Python provides
functions, variables, control flow and classes, everything you'd be
likely to need.
--
joq
6:05 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Wed, Aug 04, 2010 at 09:56:31AM -0500, Jack O'Quin wrote:
On Wed, Aug 4, 2010 at 8:01 AM,
<fons(a)kokkinizita.net> wrote:
In particular when you consider that an Ambdec configuration consists
mostly of numbers that have to be calculated and that just can't be
modified 'manually' in any way that makes sense. If you want to
experiment and modify the matrices you need programming. Using Python
means you can do the calculations as part of the configuration, and
e.g. link them to parameters set in the GUI.
More generally I always tend to view a configuration as something
imperative rather than just a description. Which is one of the reasons
why I dislike XML and friends in this role.
Ciao,
--
FA
There are three of them, and Alleline.
I'm still in doubt what to do with the
configuration files. Since adding
OSC is easy, they could just be the textual form of OSC they are now.
The other option is to skip this and turn the config files into active
Python code, directly using the API.
Integrating a Python interpreter with real-time code and event-driven
GUI code requires some trickery which I'll explore and hopefully get
working while in holiday (end of this month).
That seems like a good approach. There's a lot to be said for using a
complete programming language for configuration. Python provides
functions, variables, control flow and classes, everything you'd be
likely to need. 
6:31 p.m.
On Wed, Aug 4, 2010 at 10:56 AM, Jack O'Quin <jack.oquin(a)gmail.com> wrote:
On Wed, Aug 4, 2010 at 8:01 AM,
<fons(a)kokkinizita.net> wrote:
That seems like a good approach. There's a lot to
be said for using a
complete programming language for configuration. Python provides
functions, variables, control flow and classes, everything you'd be
likely to need.
when i did this with ardour (and a few other projects) back in the
early days, when guile was the interpreter of choice, any benefits it
brought to the table were negated by its intrusiveness into the
overall process structure. moreover, it ended up being overkill for
the purpose, and when it was removed and replaced with everybody's
bastard child (XML), nothing was lost ...
not sure that this applies to fons' planned use of python for this
purpose, or if python suffers from the same level of issues that guile
did.
--p
11:45 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Wed, Aug 04, 2010 at 12:31:49PM -0400, Paul Davis wrote:
when i did this with ardour (and a few other projects)
back in the
early days, when guile was the interpreter of choice, any benefits it
brought to the table were negated by its intrusiveness into the
overall process structure.
In what way ?
not sure that this applies to fons' planned use of
python for this
purpose, or if python suffers from the same level of issues that guile
did.
Which were ... ?
Ciao,
--
FA
There are three of them, and Alleline.
5 Aug
5 Aug
2:44 a.m.
On Wed, Aug 4, 2010 at 5:45 PM, <fons(a)kokkinizita.net> wrote:
On Wed, Aug 04, 2010 at 12:31:49PM -0400, Paul Davis
wrote:
guile wants to take ownership of the top-level exection context of the
process. i.e. in main() you have to launch/step into guile, then tell
guile to execute the "rest" of the application. this just felt all
wrong, and meant that any backtraces of the main thread ultimately
ended up back in the guile interpreter. i became suspicious about
other aspects of guile's "presence" in the process, but didn't bother
to investigate them all that much - we just dumped it for XML.
when i did this with ardour (and a few other
projects) back in the
early days, when guile was the interpreter of choice, any benefits it
brought to the table were negated by its intrusiveness into the
overall process structure.
In what way ?
4 Aug
4 Aug
3:32 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Tue, Aug 03, 2010 at 11:11:28PM -0700, Fernando Lopez-Lezcano wrote:
That would be great... right now I'm using a
HPF/LPF pair for my tests
(butterworth 12dB/oct, supercollider) and placing them high enough that,
well, maybe both the loudspeakers and the subs are in a region that is
not right where they are transitioning (~70Hz).
You are aware that those filters will be in antiphase in the xover
region ?
But I'm not really sure how I can actually measure
the bloody thing...
Measuring subs (or anything LF) is hell... unless you have some
very big space.
Ciao,
--
FA
There are three of them, and Alleline.
6:44 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Wed, 2010-08-04 at 15:32 +0200, fons(a)kokkinizita.net wrote:
Yeah, that is what it looks like...
-- Fernando
On Tue, Aug 03, 2010 at 11:11:28PM -0700, Fernando
Lopez-Lezcano wrote:
What would you recommend?
That would be great... right now I'm using a
HPF/LPF pair for my tests
(butterworth 12dB/oct, supercollider) and placing them high enough that,
well, maybe both the loudspeakers and the subs are in a region that is
not right where they are transitioning (~70Hz).
You are aware that those filters will be in antiphase in the xover
region ? But I'm
not really sure how I can actually measure the bloody thing...
Measuring subs (or anything LF) is hell... unless you have some
very big space.
7:08 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Wed, Aug 04, 2010 at 09:43:52AM -0700, Fernando Lopez-Lezcano wrote:
Invert one of the two outputs...
If you have a large outside space with a hard and flat surface (e.g.
a large parking area with no nearby buildings, or an airfield runway),
one solution is put both the speaker and the measurement mic on the
ground (for the mic that really means at most a few cm above ground).
The mic will then see two speakers, the real one and its mirror image,
and be located at roughly the same distance from both. For low
frequencies the two will combine without disrupting the response.
Ciao,
--
FA
There are three of them, and Alleline.
You are aware
that those filters will be in antiphase in the xover
region ?
What would you recommend? Measuring subs
(or anything LF) is hell... unless you have some
very big space.
Yeah, that is what it looks like...
3 Aug
3 Aug
9:45 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Tue, 2010-08-03 at 20:17 +0200, Jörn Nettingsmeier wrote:
I currently have 8 Mackie 824's at ear level, 4 624's in the ceiling and
4 624's below the floor (and I will add two more 624's, one zenith, one
nadir). So their frequency responses don't exactly match in the low
end.
Right now I'm writing a software crossover with HPF / LPF sc ugens
(12dB/oct butterworths).
On 08/03/2010 07:45 PM, Fernando Lopez-Lezcano wrote:
Yes, of course, that was my mistake - I don't have subs in the ceiling
or floor (yet :-)
Hmmm, so, you send WXYZ to the four subs and whatever order you can
decode to the regular speakers (6, 8, etc). Sounds good.
it does :) except you'd usually throw away z, What if you
want to include a crossover? (that's exactly what I'm
working on right now in the openmixer software).
So, you high pass WXYZ and send that to the regular speakers, low pass
WXYZ and send that to the subs... and what do you do with the rest of
the Ambisonics components? I imagine I would just send the full
frequency range to the regular speakers and let them do the best they
can, right?
hmm, i think i would rely on the internal electronics of the tops for
the HPF and feed them a full-range signal of all components, and then
maybe measure one and design a suitable roll-off for the bass
speakers... but i have never had the luxury to do that, i always set the
LPF by ear. sending high-passed WXYZ and full-range higher orders
to the tops won't
work well, i guess. if you want to take care of the band-splitting in
your mixer, your best bet would be ganged high/low-pass filters for all
the orders.
That is what I'm currently trying. I still don't have a clear idea of
what would be best - or even reasonable - in terms of high order
components (at this point in time I only have 2 subs so I'm sending them
just W, but the code supports 4 and I plan to have that setup soon, at
least for testing). Fons?
-- Fernando
10:54 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On 08/03/2010 09:44 PM, Fernando Lopez-Lezcano wrote:
I currently have 8 Mackie 824's at ear level, 4
624's in the ceiling and
4 624's below the floor (and I will add two more 624's, one zenith, one
nadir). So their frequency responses don't exactly match in the low
end.
in that case, you should probably measure the phase responses and make
sure they match closely, at least in the range where the ear relies on
phase cues (some sources cite up to 700hz, gerzon among them). above
that, wavelengths are getting so short (and the sweetspot so small) that
we can't rely on strictly constructive interference in most of the
listening area anyways...
i've never tried mixed settings myself, though.
best,
jörn
22 Aug
22 Aug
7:09 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Tue, 2010-08-03 at 10:45 -0700, Fernando Lopez-Lezcano wrote:
On Sun, 2010-08-01 at 13:39 +0200, Jörn Nettingsmeier
wrote:
Perhpas i missed something further down in this discussion, but what
would be the deficit to letting an outboard crossover (like the ones in
the subs) handle the passing vlf instead of setting up ambdec to do it?
On 07/31/2010 11:14 PM, "Bearcat M.
Şandor" wrote:
Hmmm, so, you send WXYZ to the four subs and whatever order you can
decode to the regular speakers (6, 8, etc). Sounds good.
What if you want to include a crossover? (that's exactly what I'm
working on right now in the openmixer software).
So, you high pass WXYZ and send that to the regular speakers, low pass
WXYZ and send that to the subs... and what do you do with the rest of
the Ambisonics components? I imagine I would just send the full
frequency range to the regular speakers and let them do the best they
can, right?
-- Fernando
I am planning on subs that have their own
low-pass filters. I have a
pair of Anthony Gallo Reference 3.1s but for this ambisonics set up, i'd
get 4 pairs of the Anthony Gallo Stradas and 4 T3 subs. stradas:
http://www.roundsound.com/reference-strada.htm t3:
http://www.roundsound.com/tr-3-subwoofers.htm Most of the stuff on
those pages is marketing of course, but the satellites have a range of
45 Hz - 20 Khz +- 3 db and the subs go down to 22 Hz (no variance given).
That push-pull idea sounds fantastic. I would love it if you'd show me
how to create a set up for a system like that using ambdec, if you have
the inclination and the time. That's what i was planning on using anyhow.
basically, you take fons' example for the octagon, set the speakers up
at the correct angles and then enter the actual distances into the
matrix, so that delay and near-field compensation is correct for your
setup. then you take the example square for the subs, set them into the
corners of the room and again enter the correct distances.
at this point you have two ambdec instances running, one for the tops,
one for the subs. you'd need to adjust the relative loudness with
ambdec's faders. once you've determined the correct relative levels, you
can factor them into the matrix coefficients of the sub decoder and run
both ambdecs at 0dB. now you can hack both matrices into one and end up
with a 12 channel ambdec configuration: 8 tops and four subs. easier to
start up in daily use.
10:40 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Sun, Aug 22, 2010 at 11:07:33AM -0600, Bearcat M. Şandor wrote:
Perhpas i missed something further down in this
discussion, but what
would be the deficit to letting an outboard crossover (like the ones in
the subs) handle the passing vlf instead of setting up ambdec to do it?
The basic problem here is that the lowpass (for the subs) and the
highpass (for the rest) have to matched in both amplitude and phase
response. If you can meausure the subs there's no problem in finding
a matching highpass. The problem is measuring the subs.
Ciao,
--
FA
There are three of them, and Alleline.
23 Aug
23 Aug
9:12 a.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Sun, 2010-08-22 at 22:40 +0200, fons(a)kokkinizita.net wrote:
On Sun, Aug 22, 2010 at 11:07:33AM -0600, Bearcat M.
Şandor wrote:
Ahh! Well i don't think matching the levels would be a problem with a
sound meter would it? (gotta get me one of those) As far as matching
phase, i have no frikin idea how you'd do that.
Perhpas i missed something further down in this
discussion, but what
would be the deficit to letting an outboard crossover (like the ones in
the subs) handle the passing vlf instead of setting up ambdec to do it?
The basic problem here is that the lowpass (for the subs) and the
highpass (for the rest) have to matched in both amplitude and phase
response. If you can meausure the subs there's no problem in finding
a matching highpass. The problem is measuring the subs.
Ciao,
9:20 p.m.
New subject: what does "full range" mean with regard to ambisonics and speakers?
On Mon, Aug 23, 2010 at 01:04:32AM -0600, Bearcat M. Şandor wrote:
Matching the level can be do just by ear. Matching filter responses
is another matter. Sound level meters are kind of useless for getting
an idea of how a sub behaves in a real room, and even a full impulse
response measurements without room effects would be quite difficult.
The only solution I see for this is room correction in the Ambisonic
domain.
--
FA
There are three of them, and Alleline.
The basic
problem here is that the lowpass (for the subs) and the
highpass (for the rest) have to matched in both amplitude and phase
response. If you can meausure the subs there's no problem in finding
a matching highpass. The problem is measuring the subs.
Ahh! Well i don't think matching the levels would be a problem with a
sound meter would it? (gotta get me one of those) As far as matching
phase, i have no frikin idea how you'd do that. 
1 Aug
1 Aug
12:34 p.m.
Hi,
On Saturday 31 July 2010 04:40:28 Bearcat M. Şandor wrote:
I'm a recovering audiophile. When i was reading
the magazines and
reading about over priced (in my now opinion) speakers, the words "full
range" tended to mean that a speaker was reasonably flat from 20 Hz to
20 kHz. Granted those were unusual.
I have read that speakers in an ambisonic set up should be "full range".
I'd like to set up a ambisonic speaker system (8 channel to start), and
the prospect of 8 full range channels is daunting. Since it seems they
would be stand or wall mounted (at least some of them) that means
monitors and subwoofers. Since all channels must be the same, that means
8 subwoofers...somewhere in the room.
So what does "full range" mean usually and what does it mean in terms of
talk on this list?
I realized that my jbl-speakers I use additionally to my alesis are a little
light on the lower end. So I (mis-)used ambdec's crossover to give me one
matrix with the 100Hz and above for the whole setup and one matrix for below
100Hz that just contains and feeds the main alesis speakers. Works pretty well
as far as I am concerned...
Of course, each speakers signal is also feed through jconvolver with a drc-ir
to make it sound superb (as opposed to just the plain good sound of the
speakers).
I think that full-range in the sense of ambisonics just means (a) that its
more then the little rear--satellites you find in 5.1/7.1 and (b) that every
speaker is supposed to get (and reproduce) the whole spectrum according to its
abilities.
Have fun,
Arnold
5234
days inactive
5257
days old
linux-audio-dev@lists.linuxaudio.org
27 comments
9 participants
participants (
participants (9)
)
-
"Bearcat M. Şandor" -
Arnold Krille -
Bearcat M. Şandor
-
Fernando Lopez-Lezcano -
Fons Adriaensen -
fons＠kokkinizita.net
-
Jack O'Quin -
Jörn Nettingsmeier -
Paul Davis
-
"Bearcat M. Şandor"
-
Arnold Krille
-
Bearcat M. Şandor
-
Fernando Lopez-Lezcano
-
Fons Adriaensen
-
fons＠kokkinizita.net
-
Jack O'Quin
-
Jörn Nettingsmeier
-
Paul Davis