I did say very explicitly transient time differences.
Maybe I am missing something but these are not detected
in Blumlein and I am not clear on why they would
be in first order. I don't think this works.
I could be wrong, however. But I think that
in Blumlein stereo anyway, everyone agrees it does
not. so that et what is called the "Glockenspiel effect"
arises where lower and higher frequencies are separated
in perceived position.
If this is wrong, I would surely be interested to know
why, and if it is what happens in BLumlein I do
not see why it would go away with first order Ambisonics
Robert
On Mon, 2 Apr 2012, Eric Benjamin wrote:
Robert,
Lots to comment on here. I seem to be compelled to address your negative or
"not so good" observations:
Not so Good 2) Because one- point miking ignores transient time
of arrival differences as such , one of the basic cues of sonic perception
is suppressed explicitly
That's not really true. I'm assuming that when you speak of time of arrival
differences that you are referring to ITDs. The thing to remember here is that
ITDs are a function of our presence in the acoustic field, and as such aren't
present in the recording environment and thus shouldn't be recorded. In a
recording and reproduction scenario the ITDs happen in the reproduction of the
recording, and as it happens ITDs are reproduced very well by Ambisonics, even
first order Ambisonics. I showed this quite clearly (I hope) in AES preprint
8242.
3) Impractical number of speakers needed really to work
But one of the really cool things about Ambisonics is that it scales extremely
well so that it works well with one speaker or two, although not creating
surround with so few speakers. And it works quite well with only four
speakers. And nowadays there are quite good decoders that work well with ITU
5-channel arrays. If higher order sources are available then they can be
decoded in such a way that the directional resolution is high in the forward
direction where there are relatively many loudspeakers and not so well to the
rear where there are relatively few loudspeakers.
4) Impractical number of channels needed to really work
Again, that's not really true. Most common audio carriers have the capability
to carry many channels, DVD, BluRay. And many systems are file-based and as
such aren't really limited at all. With a system that is inherently
hierarchical, as Ambisonics is, a broadcast or distrubution system can transmit
as many or as few channels as is wished.
5) In practice, keeping noise low enough is difficult
I'm not entirely sure where this comment comes from. In terms of natural
recording, which is what you and I would do but not most of the rest of the
audio world, the Soundfield microphone as embodied in the Soundfield MkIV and
MkV microphones, is really quite quiet. Not as quiet as some modern
microphones, some of which have self-noise in single digits, but somewhere in
the mid-teens of dB SPL. I can't bring to mind any instance when listening to
the recordings of my colleague Aaron Heller that I was ever aware of the
presence of noise. And there's no reason why higher order systems can't be made
very quiet indeed. Gary Elko mentioned, during the discussion of the MH
acoustics Eigenmike, that the self noise of the zero order (omni) output is
about 0 dBA.
So what does my list look like?
Good:
1) Isotropic behavior. Ambisonics is really good at capturing and reproducing
ambient sounds. These are the sounds that inform me that the sound scene had
some real origin.
2) Reproduction of correct timbre. While it is relatively easy (but not
frequently done!) to capture sound with the correct spectrum, 2-channel stereo
distorts that spectral accuracy in reproduction. Ambisonics is much better
although it still suffers from some of the same problems.
3) Requires lots less speakers than Wave Field Synthesis.
Not so good:
1) I frequently find that I have front/back confusion.
Let the debate continue.
----- Original Message ----
From: Robert Greene <gre...@math.ucla.edu>
To: Surround Sound discussion group <sursound@music.vt.edu>
Sent: Sun, April 1, 2012 8:03:44 PM
Subject: Re: [Sursound] Can anyone help with my dissertation please?
OK I thought that was a good idea, for people to say what they thought
was good and not good about Ambisonics. So here I go(first I guess
but my mother always said Act in haste, repent at leisure. I think
she meant it as cautionary but I have always taken it as advisory!).
Good
1 Elegant as mathematics
2 Forces people to use one point miking which in itself
is already a HUGE thing because it eliminates the absurd
manipulativeness of much of commercial recording practice.
3 In principle, has the capability of reconstructing the complete
soundfield.
4 Puts height in the picture and gets rid of the sound through
a horizontal slit of stereo(which is ironically more like that the better it is
done!)
5 In practice, more robust than one might have expected
at working over a large listening area (if that matters).
6 In principle, the timbre errors of stereo arising from around the head
summation are eliminated.
Not so good
1 Emphasis on homogeneity makes it inefficient when not high order.
(Everyone knows that perception to the side of a listener is quite different
from perception frontally, but this is ignored)
2 (related to 1) Because one- point miking ignores transient time
of arrival differences as such , one of the basic cues of sonic perception
is suppressed explicitly and is only returned to the picture with higher order.
3 Impractical number of speakers needed really to work(cf point 2).
4 Impractical number of channels needed to really work(because
higher order is needed).
5 In practice, keeping noise low enough is difficult.
6 Nearly oomplete lack of demo material, which makes it all
but impossible to interest the public.
One somewhat incidental issue
7 Mathematics is too tricky for most people in audio to appreciate
(I know this is so because I have tried to write about Ambisonics
for the general audio public--no dice, people did not get it even
though I thought what I wrote was clear as crystal)
About point 2: the same issue arises in Blumlein stereo, which is why some
people like ORTF better. Ideal would be Blumlein up to around 700 Hz and switch
to ORTF above that, it seems, or something along those lines.
(Blumlein the man had ideas on this, of course).
I could go on, but perhaps that is enough to get the ball rolling.
Good suggestion, I think, that people should make such lists. I am
curious to see what others have to say.
Robert
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound
