Talk:Ambisonics/Archive 1

John Hayes & Duane Cooper
I'm reasonably familiar with the Ambisonics literature, but am not familiar with John Hayes' contribution. Could someone fill me in? Duane Cooper should probably be listed as an inventor well. My understanding is that Cooper, Fellgett, and Gerzon each independently came up with the idea of using spherical harmonics to represent the sound field.--AJH 21:54, 15 October 2006 (UTC)

Duane Cooper was one of the developers of Nippon Columbia's UD-4 quad system. The UHJ system of coding Ambisonics relies on some aspects of his work and on work done by the BBC on Matrix H. The "U" ("Universal") in UHJ is apparently a reference to UD-4. I've added this to my contribution on UHJ on this page. John Hayes was at speaker manufacturer IMF, which released a very early decoder and was involved in aspects of the original Soundfield microphone. The company also released some very early 2-channel Ambisonic LPs. see http://imf-electronics.com/Ambisonic/index.html Richard E 16:34, 7 January 2007 (UTC)


 * Should John Hayes be listed as an inventor of ambisonics at all? His own account given at http://imf-electronics.com/history/index.html doesn't seem to justify it. Paul Hodges 14:57, 14 January 2007 (UTC)


 * As he says, "We formed a team, myself, John Wright, David Brown, Prof. Felgett and Michael Gurzon (sic). I remember spending many hours in the anacoic chamber working on the surround sound microphone..." then I would say, definitely, yes he should. However, one could reasonably ask that if Hayes is in, then Wright should also be in. I never knew David Brown, but essentially the above listed names constitute the original Ambisonic research group. I'll look at it. Richard E 15:08, 14 January 2007 (UTC)


 * The difficulty is in deciding at what point the act of invention takes place. As Ambisonics can exist without reference to the Soundfield microphone, is working on the microphone part of the invention?  I'm not trying to do anyone down - just to try and get a clearer picture of who did what when, as it were. Paul Hodges 15:11, 14 January 2007 (UTC)


 * I've tweaked the list of people at the top of the article. See what you think, and/or tweak as appropriate.Richard E 15:18, 14 January 2007 (UTC).


 * I think this is bound to be a bit arbitrary. I've now listed the complete original Ambisonic team, which did a good deal more than developing the SFM of course. The way I've written it slightly credits Gerzon with the original work (probably accurately) but only slightly. Fellgett has to come next, and after that you have the IMF guys who did all the original dev hardware and early production stuff (such as a decoder) before Calrec came on the scene and produced production gear. The "second wave" of researchers includes Geoff Barton, who came in via being at Reading under Fellgett, and I suppose some might consider me (and Pete Carbines and George Chkiantz, aka the "Ambisonic Mixing Group") as part of the second wave too - we were certainly doing mixing research, implementing Gerzon's original panpots etc, very early on. Geoff did originally get a credit in the main article but he may now only appear in the UHJ article. Richard E 15:28, 14 January 2007 (UTC)


 * That looks good to me, and fits my (faint) memories of conversations with Michael at the time. Paul Hodges 17:13, 14 January 2007 (UTC)

G-Format
Hi there - why is it ever impossible to decode G-Format to horizontal-only B-Format? Didn't want to change the page because I might be missing something, but... why? Cheers :-)


 * In two words, "shelf filters". The person creating the G-format can decide to apply shelf filters beforehand, and doing so will improve the listeners' experience. Unfortuately, shelf filters are not standardised, so how do you remove them to get back to B-format?  (If they were standardised then it would be easy.)  Martin.leese 04:55, 9 January 2007 (UTC)


 * It is possible to envisage a G-Format structure that allowed for height information, but one does not currently exist. G-Format is "defined" (if that's not too strong a term) as decoding a B-Format signal to a "conventional" surround array in the studio, and distributing the resulting speaker feeds via a conventional multichannel medium. While there are agreed standards for planar surround (5.1, 7.1, 10.2, whatever nonsense you want to come up with in the ultra-inefficient world of one channel:one speaker), there is not a standard for carrying height information in a conventional surround system. Several people and companies have suggested or used methods of doing this with conventional 5.1 layouts, primarily by reallocating arguably less-important (CF) or unnecessary (LFE) channels to carry height information: Chesky reallocates the 5.1 CF and LFE as a pair of elevated speakers halfway down the array; Telarc has used the LFE as an overhead. MDG and some others have used "2+2+2" which includes a height pair at the front (why the front? Don't ask me). While there is a special 2+2+2 flag in DVD-Audio (and thus presumably in other MLP-based systems on HD discs, ie TrueHD), there is not a "standard" for including height information. One might emerge in the context of HD disc formats and if so, one could decode 4-channel with-height B-Format to a G-Format that included height capability. Richard E 09:53, 9 January 2007 (UTC)


 * And to finally answer the question: Were one to use a system like that described above and you didn't use shelf-filters in the decode (or you used shelf-filters that were in some way reversible - I have had mixed messages on this, some say you can do it, others that you can't - comments?), then you could recover periphonic B-format from full-sphere G-Format. Richard E 13:08, 12 January 2007 (UTC)

G-Format with height?
The section below was written by me in answer to the question above. Martin has queried its necessity and is probably right. It's here in case anyone wants to review it and make any suggestions. It is entirely possible to create G-Format recordings that include height information. However, while there are "standards" for conventional planar surround (5.1, 7.1 etc) there is currently no recognised standard for the inclusion of height. This being said, there are several techniques around, the most common taking one or two channels of the 5.1 signal (typically LFE, or CF & LFE) and using them to drive elevated loudspeaker(s). It would be possible to decode an Ambisonic with-height recording to configurations like this and release the result as described. Richard E 10:42, 10 January 2007 (UTC)

Am putting this back under "Current developments". Martin.leese 23:42, 10 January 2007 (UTC)

Hardware & Software & The Current Age
Might it be worth updating the page a little to mention some of the software that's out there? The page reads as if all Ambisonics is done on valve amps. I'm pretty sure more Ambisonic encoding and decoding is done in software these days than any other way. I've about 40GB of B-Format and FMH material that's never been anywhere near a tape player, let alone a "traditional" "2-channel disc" (what's one of these?). The history is cool - perhaps we should reorganise the page a little so there's a "then" and "now" section?


 * I have partially addresed this by adding a new section "Downloadable B-Format files". Martin.leese 08:14, 9 January 2007 (UTC)

Also, it might be worth mentioning explicitly how Ambisonics differs from other surround sound techniques and some cases where famous musicians have used it (I've a feeling both Orbital and Pink Floyd have played with it, but that could be apocryphal).


 * A "traditional 2-channel disc" includes something like a Compact Disc, still the most popular music distribution format. :-) However the point is valid. I've been meaning to write up a section on plugins etc but have not had the chance yet. They are mentioned (twice) but not expanded upon. Richard E 09:45, 9 January 2007 (UTC)

Ambisonic Surround Sound FAQ
I created and maintain the Ambisonic Surround Sound FAQ. I added this to the "External links" section of the Ambisonics article, but an editor has decided this is a conflict of interest. This is because I maintain the Web page on which the FAQ sits.

The Ambisonic Surround Sound FAQ is a major resource for Ambisonics; it should be, most of the people who created Ambisonics have contributed to it. (They are listed in Section 24 of the FAQ.) I think that the Ambisonics FAQ deserves to be on the Ambisonics page. Please discuss this (and then somebody do it). Martin.leese 08:09, 7 January 2007 (UTC)

I completely agree with Martin. Before I brought my Ambisonic.net site on-line, and to this day, Martin's FAQ makes a valuable central contribution to the documentation of this system. I actually added it before reading this page, but hopefully nobody disagrees! Richard E 16:39, 7 January 2007 (UTC)

Ambiguity?
I don't follow the equation following "More generally, the full three dimensional case is given by:...". What on Earth are the angles and elevations here supposed to be? The equation makes no sense without this context (I'm guessing equal spacing over the sphere?). I nearly just removed the paragraph, but maybe there's just some text missing. (?) —The preceding unsigned comment was added by 80.195.230.226 (talk) 02:09, 9 January 2007 (UTC). (Hmm, twas I, Richard F - need to work out how to sign this stuff.)
 * I'm guessing equal spacing on a sphere too. I will put it in and, if it's wrong, somebody can replace it.  Martin.leese 04:58, 9 January 2007 (UTC)


 * Later. I question the value of including decoding equations at all.  Different decoder designs -- there are at least three -- use different equations.  Over on the sursound e-mailing list they think these particular equations are for a "diametric decoding" design.  Martin.leese 05:56, 9 January 2007 (UTC)


 * Perhaps we need a full section on decoding. Note that this equation was put in by me ages ago and is basically the simplest decode you can "get away" with. I should have been more explicit about that but I seem to remember that it was all done in a rush when I found the original Wikipedia Ambisonics entry - which at the time was not particularly good. I'm not sure why there is confusion about the terms in the equation, since it does say that "where N is the number of speakers, n is the speaker under consideration, k is a constant which can be between 1 and 2 and θn is the angle at which the speaker is located on the circle." for the horizontal case and "where φn is the angle above or below the horizontal plane." after the with height equation --DGMatic 12:58, 9 January 2007 (UTC)

Martin Leese should be given the opportunity to create a comprehensive Ambisonics article
I was an enthusiastic Ambisonics supporter for almost two decades, and contributed a three-part article to Audio Amateur on the subject (which Peter Fellgett thought well of). I've also made my own B-format recordings, using both discrete mics and the SoundField mic.

Martin Leese has invested a huge amount of time in creating and maintaining the Ambisonic FAQ material. Making it available through Wikipedia would not only enhance Wikipedia with an article on one of the most important advances in sound recording and reproduction since Edison's phonograph, but it would make knowledge of this technology widely available.

Though Martin and I have mildly disagreed over a few philosophical points, I've found his FAQ material to be both correct and comprehensive. As Martin almost always knows what he's talking about, and has no financial interest in Ambisonics (that I'm aware of), I see no conflict of interest.

--WilliamSommerwerck 11:46, 7 January 2007 (UTC)


 * Thank you for your kind words, William. The essence of Wikipedia is that articles are community efforts, so I would not wish to create the article.  For good or bad, it must be a joint effort.  I am happy now that Richard E has included the FAQ as an external link.  This will give readers access to my work.  Martin.leese 23:39, 7 January 2007 (UTC)

To my knowledge, nobody has any direct financial interest in Ambisonics these days with the exception of Dr Geoffrey Barton and his Trifield system (a 3-speaker stereo rendering system based on Ambisonics). Apart from this and the so-called "Vienna" decoder design for decoding to irregular arrays, virtually all Ambisonic IP is in the public domain.

My employers, Meridian Audio Ltd, include Ambisonic decoding in their surround processors but its presence or absence doesn't impact sales.

Richard E 16:39, 7 January 2007 (UTC)

FMH-Format section
I deleted the following section:

==FMH-Format== FMH-Format is a "second order" format for Ambisonics, using nine channels rather than the four used by B-Format. Significantly more spatial information is captured in this format. At present, "real" recording techniques using this are in their infancy, however it is straightforward to compose synthetic recordings.

There is no format called "FMH-format". The section seemed to describe second-order Ambisonics; this is still called "B-format", same as first-order. Also, higher-order systems are mentioned in the Ambisonics section.

Martin.leese 05:38, 9 January 2007 (UTC)


 * Later. It now occurs to me that the poster may have been thinking of the Furse-Malham Higher Order Format.  However, this is a set of coefficients for creating second-order B-format channels.  A file format for downloadable B-Format files now exists, and this is defined for upto third-order B-format, so clearly second-order B-format does not have a special name.  (The only reason we didn't go higher than third-order is because the number of channels is used to uniquely define the order, and after third-order ambiguities creep in.  For fourth-order and above, a new file format will be needed.)  Martin.leese 07:44, 9 January 2007 (UTC)


 * I think we can cobble this into something we can put back. Stand by... Richard E 10:26, 9 January 2007 (UTC)


 * Speaking as one of the originators of the term FMH, which does indeed stand for Furse-Malham Higher order format, this was (is) applied to the 2nd order format that Richard Furse and I derived back in '99 (see ). Its use is intended to indicate that weightings were applied to the channels such that all the coefficients have a maximum value of 1.0 (or minus 1). This makes them not quite kosher in mathematical terms but better in engineering terms. FMH in turn was a development ( read "fix") of the strange hybrid variant that appeared in my 1999 ICMC paper. Richard and I thrashed it out between the submission of my paper for publication and the actual conference so I was able to announce it as a correction at the conference. I, personally, am not particularly bothered if the term vanishes into obscurity but I thought you might like to know where it came from.
 * Although the published version was only 2nd order there was an implication that this normalisation to 1 (apart from W's 0.707) was to be continued to higher orders.--DGMatic 12:33, 9 January 2007 (UTC)


 * Ah, interesting. Now I look at the pages, I can see that the "FMH" term isn't particularly prominent, but it is used and does work - I've pieces by a few people in the format. It may be a bit obscure for presentation - I'll leave that with other folk! (Richard F)

Obviously I was a little heavy handed, but I don't believe we want different names for B-Format at different orders. The whole business of higher orders is complicated enough as it is. The FM set is second-order as far as everyone is concerned. Any chance of the second-order pieces being made available on Ambisonic Bootlegs? I haven't checked all 75 pieces there, but suspect that they are all first-order. Martin.leese 02:36, 10 January 2007 (UTC)


 * Also, should we be dictating the use of the Furse-Malham set in the ".amb" specifiation. In a sense B-Format is an an abstract concept, but the ".amb" file format isn't; that's concrete.  If I have understood things correctly, we would also have to define what to do with the W channel for the mixed-order cases (6, 8, and possibly 11 channels).  Martin.leese 02:27, 11 January 2007 (UTC)

Richard F here: why is the Furse-Malham stuff not correct mathematically? It's normalised for single signal peak and is defined in a mathematically rigorous way - you'd have problems with fixed noise floor transmission if you were using orthonormal versions of the functions. Could we just mention that it's normalised for transmission? Cheers...


 * Blame Dave M, this was his terminology :-) Change the article, I am sure nobody will object. By the way, any chance of your second-order pieces being placed on Ambisonia.com? At the moment, all the stuff there is first-order. Some second-order stuff would give encouragement to player developers to incorporate secord-order decoding. HairyWombat (talk) 15:05, 18 March 2008 (UTC) [Actually Martin L]

Description of W
Is not the following:

"The W channel is the general mono level of the signal, corresponding to left + right + front + back (+ up + down where present)."

more-or-less gibberish? It would surely be simpler to say "The W channel is a non-directional mono signal (as output by a perfect omnidirectional microphone)", and leave out the stuff about directions - which in any case could be taken to imply that W is different if Z is absent! Similarly, the X, Y, Z components could be additionally described as like the outputs of three perfect figure-of-8 microphones.

These changes would help relate the signals more directly to people's experience, and would form a basis for a mention of "native" B-format microphones such as the Halliday-Nimbus, if this was felt appropriate (I was surprised not to see one).

Paul Hodges 11:35, 9 January 2007 (UTC)


 * I don't think it's gibberish (obviously - I wrote it! :-)) but I am happy if someone wants to rewrite it better - I'm trying to cover a lot of ground on this at present and I'm not focusing on all of it at the same time.


 * The reason I added the sum and difference signal descriptions is that they are really important when it comes to understanding the relationship with stereo techniques based on the same principles.


 * And yes, W is effectively different if Z is absent! You could have a 'native' mic that included a microphone that was only omnidirectional in the horizontal plane and had no vertical pickup at all, but this would not work if you needed height.


 * I've now tweaked this and I hope people are happy with it. If not, please feel free to tweak it yourself! Comments and suggestions welcome. Richard E

I prefer your revised wording, though I'm still struggling to find any purpose to the description of the signals in terms of some undefined directional components. I guess you're building them up from a notional set of six cardioid mics, but why this is more useful than the other description escapes me! Paul Hodges 13:33, 9 January 2007 (UTC)


 * Well, no, I'm building it from a notional set of three figure-8s and an omni! Perhaps I like the sum and difference description because the sum-and-difference-like-Blumlein was the original description of B-Format that I was given, I suspect by Gerzon (who also gave me a mathematical description but the sum-and-difference version is more use to recording engineers). It's aways struck me as simple, elegant and easy to understand. Anyway, it's there for people who find it useful, and now hopefully does not obscure alternative explanations that might be more useful to others. Richard E 13:47, 9 January 2007 (UTC)

I'm surprised if recording engineers would find "front + back + left + right + up + down" a useful definition of omnidirectional and "front - back" a useful definition of figure-8, as these are terms they are already familiar with. But I'm sure you know more of them than I do. I can see possible value in relating the signals W+X and Y to M and S in an MS microphone setup; but this should perhaps not be right at the point of the initial definition of B-format. Paul Hodges 14:02, 9 January 2007 (UTC)


 * At the risk of circular reasoning, I would say that the average recording engineer is certainly aware of the nature of omni and figure-8 mics in practical terms: they will also be familiar with the idea that coincident stereo recording employs sum and difference signals. The "traditional" omni coincident with a left-facing figure-8 yields two signals, L+R and L-R respectively. Obviously, these are matrixed together to generate left and right channels by adding them in-phase: (L+R)+(L-R)=2L and out of phase: (L+R)-(L-R)=2R (I'm including these in case other readers don't know them). I believe the knowledge that B-Format works the same way in three dimensions and its implicit link back to Blumlein to be helpful. They may also know of the same technique formerly used in some broadcast recording systems to minimise the effects of tape azimuth variations. Meanwhile, to most recording (as opposed to audio) engineers, the concept of describing a soundfield in terms of spherical harmonics remains, I am sure, utterly alien. Richard E 14:40, 9 January 2007 (UTC)

While you were writing that, I've removed the part I think adds nothing, and added a separately headed sub-paragraph relating coincident techniques to B-format - I'm sure you'll say if you don't approve. My intention was to remove the complication of relating the B-format signals to the familiar ones from the "pure" description of them. Paul Hodges 14:45, 9 January 2007 (UTC)


 * No problem with the new section, but I really would like to retain the sum-and-difference description of B-Format. I've reinstated it in shortened form and in your new Coincident Stereo section (which I think adds useful information but does not cover the ground my description did previously). Hopefully you will not have any further difficulty with the inclusion of this information, especially as it is now separated from the initial introduction of B-Format. Richard E 14:51, 9 January 2007 (UTC)

I think that's fine now. Paul Hodges 14:54, 9 January 2007 (UTC)


 * Phew! Thanks for the contribs, I think they've improved it significantly! Sorry it took some effort to get there... Richard E 14:56, 9 January 2007 (UTC)


 * Not to ruin the harmony here, but IMHO there is a 90 degree phase error in the current description. In a plane wave, pressure and velocity are 90 degrees out of phase, whereas the W channel will be in phase (or in opposite phase) with the three other channels. A more precise description would perhaps be to say that the W channel is the time derivative of sound pressure (multiplied by constant), whereas X, Y and Z are space derivatives. That has a nice time-space-symmetry to it as well.


 * Sveinb (talk) 11:53, 14 May 2009 (UTC)

Don't confuse velocity and pressure gradient - that's where your phase error has come from (perhaps we should downplay or qualify the mention of velocity). Also, don't forget that the W signal is not directly connected to the X, Y, Z - if it was, there would be no point in recording it. Paul Hodges (talk) 09:57, 17 May 2009 (UTC)

Higher Order
In the PDF linked at http://www.york.ac.uk/inst/mustech/3d_audio/higher_order_ambisonics.pdf under the higher order section, the rotation matrice are wrong. The author Dave Malham has confired the type errors to me a few month over email. I think he was intending to regenerate the PDF file from the original text to fix the type error but it seems he did not get to it yet. Dave, perhaps you could put the correct version up when you can? Aristotel

Guidelines for editors?
Would it be useful to have a section on the Talk page giving guidance?
 * As long as it is Ambisonics-based. There should be no need for general editorial guidelines for which Wikipedia standards already exist. Richard E

Given the lack of agreement, the answer would seem to be "no".

What I have in mind is stuff like:
 * Avoid the term "3-D" because many people thought that two-speaker stereo was 2-D, and that 5.1 is 3-D.
 * I must say, I've never heard that misconception before. It seems completely ridiculous. Richard E
 * Completely ridiculous, but very common. I have been hearing it for years on newsgroups.  Here is an recent expample from the Talk page for Ultra High Definition Video (with 22.2 audio):  Martin.leese 19:46, 11 January 2007 (UTC)

 "24 speakers? Ugh. Why should we use 24 speakers? I think that 5 speakers is more than enough to produce 3D audio. Henri Tapani Heinonen 11:01, 7 November 2005 (UTC)" 
 * --sigh--... I take your point. (the correct answer is: "7 speakers are more than enough to produce 3D audio" :-)


 * Use "full-sphere" in preference to "with height" because the former implies (correctly) the reproduction of a spherical soundfield, while the latter implies a couple of speakers stuck up in the air somewhere.
 * I don't agree with the latter. And the former suffers from an unavoidable tendency (as does "periphony"), to suggest that you can only localise sources at the surface of a sphere and not within it. Richard E
 * Then we need a new term. Martin.leese 19:46, 11 January 2007 (UTC)
 * I think the current terms are sufficiently well understood. I would rather define them more rigorously in the context of the article than come up with new ones. I have attempted to do this in the Notes on Nomenclature section. Does that do the job? Certainly, I would not want to replace "periphony" for example. Richard E 12:17, 12 January 2007 (UTC)
 * Actually, thinking about it more deeply, there's nothing about "full-sphere" which infers "only on the surface of the sphere". Please forgive my BS.Richard E 12:55, 12 January 2007 (UTC)


 * I have "depersonalised" the section on terminology - not that I prefer passive writing, but it sounded inappropriately clubbish saying "we use...". And I added the term pantophonic as quite a lot of references use it. Paul Hodges 12:40, 12 January 2007 (UTC)
 * Much as I also dislike the passive, I think this is an improvement. I think the add of "pantophonic" is a bit ungainly, but I can't think of a better way of putting it at present. Richard E 13:00, 12 January 2007 (UTC)


 * There will be two types of reader, General Readers and Specialised Readers. Stuff sections for the General Reader near the top, and specialised sections near the bottom.  Do the same within sections.
 * I certainly think that you should get an introduction to a subject at the top. It may not always be practical to do this in subsections, however. Richard E


 * Hit "Enter" occasionally. This has no effect on display and makes diffs in the History not so wide.
 * Actually, create para breaks frequently. Visual breaks in the text help reading immensely, especially as we are on a site with one column, small font size and as a result a large number of chars/line. Richard E 12:22, 11 January 2007 (UTC)
 * You misunderstand. I meant hit "Enter" within paragraphs to give 80-character or so line lengths.  The Wiki display system ignores these (as do browsers), but they make life much easier when doing History diffs.  (I agree that paragraph breaks are also important.)  Martin.leese 23:13, 12 January 2007 (UTC)

Can't think of any more at the moment. Martin.leese 02:40, 11 January 2007 (UTC)
 * Confine active research topics within "Current discussion".

Trifield
I considered giving Trifield its own section but decided against it as the process relies so much on Ambisonics that it can be considered a special case. Looking at its own stub as it now stands, I think it was better in the main body of the article. However I do not think that strongly enough to want to undo Martin's decision to move it. Observations welcome.Richard E 12:11, 12 January 2007 (UTC)
 * One reason for carving Trifield out into its own article is to reduce the length of this article. Whether Trifield is Ambisonics is debatable.  A major difference is that Ambisonics aims to satisfy localisation cues above and below 700 Hz, whereas Trifield aims to satisfy cues above and below 5 kHz.  Martin.leese 23:26, 12 January 2007 (UTC)

Meridian Audio Limited
As an employee of Meridian I was unwilling to add links or references to the company. Now these have been added it might wish to be noted that in fact Meridian was an early licensee and currently offers a number of surround processors/controllers that include Ambisonic decode capability, which is standard in Meridian surround products.
 * Will add "standard in Meridian surround products". That's useful info.  Martin.leese 23:20, 12 January 2007 (UTC)

Meridian decoders handle both horizontal B-Format and 2-channel UHJ and provide decoding to all arrays supported by the decoder (typically up to 7.1 and beyond). The decoders also support Super-Stereo and Trifield decoding which incorporate a subset of the same code. Current products with Ambisonic and Trifield capability include the G61, G68, G91A, G95, C61, and 861. More details are available on the Meridian web site. Richard E 12:11, 12 January 2007 (UTC)
 * This might be more appropriate in the Wikipedia article on Meridian. Also, Meridian can't be accused of advertising when you add stuff to an article about you.  Martin.leese 23:20, 12 January 2007 (UTC)

National varieties of English
There have been two main contributors to this article so far, one British and the other American. It would appear that in all but one case (the spelling of "centre", where one occurrence was American and the rest English) the article used British English spellings where the two were different. It is purely arbitrary to determine that henceforth the article will be in US English (and alter it accordingly), and this should not be done without discussion.

As I have written in both US and British English for years, I have no trouble working in either. However, I can imagine that a number of future UK-based contributors and readers may be rather upset that the article on this British invention, in which all but one of the significant developers was also British, is now written in US English when it previously was not.

There is no doubt that the article has to be written in one consistent variety of English (Manual of Style (national varieties of English)), and I would propose that as this is a British invention, described in an article here that was initially written almost entirely in British English, with virtually all references and articles on the topic written in British English, that British English be used for said article. Comments please - I am not reverting it without discussion. Richard E 01:24, 14 January 2007 (UTC)


 * I am reverting it. I read a different part of the Wikipedia guidelines (which I can't find now) which suggested that articles should follow American English.  As there is no need for this, we shouldn't do it.  Martin.leese (Manx by birth, Canadian by choice) 03:42, 14 January 2007 (UTC)


 * Aha! I didn't know you were Manx! My apologies for any inadvertent incorrect assumptions. The use of English variants in Canada is very interesting. My wife is Canadian and a Commonwealth Lit professor so this comes up quite a lot. Many staff in Canadian universities are British and while there is obviously a popular cultural pressure for US English, both variants are well-known, acceptable and available to Canadians. So, for example, she and I just had a discussion about which pronounciation of "Renaissance" was "correct" in Canada: quite a lot of people use the US "REN-iss-aunce" but the English "Re-NAY-sonce" is perfectly acceptable. Many US English spelling variants were cooked up by Noah Webster, who travelled the US encouraging printers and publishers to spell words differently to the English way as a pro-Independence statement. In Canada you can avoid one sort of cultural imperialism - as long as you endorse the other. :-) I'll go a-hunting for the reference about using US English myself - when I saw you had Americanized the spelling I thought that you might have found a Wikipedia policy that suggested it (the English Wikipedia is hosted on US servers and run by a US corporation, after all) but instead I found the above. Having been a US resident for 12 years I am happy with and used to working in either medium: my policy when editing articles for international journals has always been to leave them in the variant in which they were written - we decided this at the Climate Action Network and it always seemed to work fine. In addition, languages have to grow and evolve and that means taking on board new usages, spellings and words. I am very glad there isn't an English Academy! Richard E 11:10, 14 January 2007 (UTC)

Decoding section
I am not sure what to do about the Decoding section. It was me who added the three sub-sections on the different decoder types. Obviously I hoped that somebody would complete the other two. I also thought that we could wait until the rest of the article was in good shape and then invite DaveM to do this. (I believe the stuff I put into "Diametric decoding" is all his.)

However, I notice that when I now edit the whole thing, it gives me a warning about its length. I would suggest, therefore, that to limit the length we delete the existing equations and combine the three sub-sections into the main "Decoding" section. This is unfortunate as DaveM obviously spent a lot of time on the equations. On the other hand, the thing is getting very long.

Anyway, let me know what you think. Martin.leese 04:04, 14 January 2007 (UTC)


 * I agree. The equations are not of the essence here.  A summary of the different decoding types, with some notes on their different effects would be more useful.  In fact, is it not less a matter of different types in most cases than of a variation in the balance of parameters?  Paul Hodges 10:24, 14 January 2007 (UTC)


 * See my item below on Article Size. I do not think trashing hard work is a solution to excessive length. The equations are of limited interest to me, but there are mathematicians in the surround community who may feel that the math validates the system, etc. A summary style article with current subsections turned into articles may be the answer.Richard E 12:31, 14 January 2007 (UTC) —The preceding unsigned comment was added by Relen (talk • contribs) 12:29, 14 January 2007 (UTC).

Article size
Given that the article is getting big, do we need such a large section on UHJ? In reality only the 2-channel form has ever been used, or ever will be; the fact that there was a hierarchy has a passing historical interest (especially the 2 1/2 channel FM experiments), but doesn't need the detail that is currently on show. Paul Hodges 10:33, 14 January 2007 (UTC)


 * There is nothing actually wrong with the section on UHJ in my view. Yes, it is of historical importance - but so is the vast majority of Wikipedia content. Yes, only 2- and 2½-channel UHJ have ever been used practically; however 2-channel UHJ is of current interest due to at least one series of current record releases using 2-channel UHJ as a source of G-Format. It is not helpful, I would suggest, to remove the background to UHJ as a whole as it is necessary to understanding the context of the 2-channel variant which is in current use. In addition, there is valid content there which has been written: consigning it to the bin is a little demoralising for contributors (me in this case) who have put a lot of effort into it.


 * It would be possible to summarise UHJ and link to a new article containing the existing UHJ content, more or less as has been done with the Trifield material - I will have a look at that possibility. Indeed the entire article could be turned into a summary with separate pages taking the place of each section, with a bit of thought. See Summary_style. Note also that the 32KB nominal size-warning limit is designed to limit readable content to this length and is "no longer a binding rule". Markup (eg the tables and equations) can cause an article to exceed 32KB where the readable content in fact does not. See Help:Page_size. Richard E 12:37, 14 January 2007 (UTC)


 * Pulled UHJ section into its own page and included summary and link at previous location. Anyone want to do this with the decode equations? Richard E 13:13, 14 January 2007 (UTC)

That's exactly what I had in mind. I didn't mean to be demoralising - The UHJ section was excellent, and preserving it in a separate article seems a very good solution. Paul Hodges 13:24, 14 January 2007 (UTC)

G format section
I was looking over the G-format section and the examples of records which come in G-format may cross the line into being an advertising drop in. Does anyone else have an opinion on this? --D 00:58, 3 May 2007 (UTC)


 * At present such records are rare and not easy to find or identify without pointers. Providing pointers enables people to investigate ambisonics G-format material for themselves if they wish. No specific Nimbus disks are mentioned, as they publish the information themselves; a specific Chesky disk is mentioned because most Chesky disks are not of this format - however no competitive or comparitive statement is made about the artistic content. Paul Hodges 13:20, 3 May 2007 (UTC)


 * As the section lists all G-Format recordings commercially released, it is just informative. If it only listed a selection then I can see how it could be thought to be advertising. Also, if there were more than ten, somebody would probably compile a list which could then be included in the External links, like the Ambisonic Discography for UHJ format recordings (of which there are several hundred). 216.123.197.21 03:55, 4 May 2007 (UTC)

tag
About a week ago the article was tagged with. Despite several requests made on the tagger's Talk page (please see User_talk:Semitransgenic), they have not said what statements they feel need verification. I have already gone through the article and referenced every statement that I felt might be challenged. As far as I can tell, the article complies with WP:VER and does not need additional references. I don't know what the tagger expects me to do. I must assume the article was tagged in error, so am deleting the tag.

If anyone can see statements that need a reference, will they please indicate them on this Talk page or, better still, with tags in the article. I will be happy to then dig out citations. However, a blanket tag applied to the whole article doesn't really help. HairyWombat (talk) 17:22, 9 January 2009 (UTC)


 * Frankly, I can't see any need for a tag on this article. Statements requiring a reference should be indicated in the text as described above and dealt with individually. Richard E (talk) 09:28, 17 May 2009 (UTC)

First-order Ambisonics and B-Format
As I have mentioned earlier, the current description must be incorrect. The W channel can not represent pressure, since pressure is 90 degrees out of phase with velocity in a plane wave, whereas an ambisonics recording of a plane wave has a W channel 0 or 180 degrees out of phase with the other channels. I believe a correct definition is that the W channel is proportional to the time derivative of pressure.


 * I am not aware of an error here. The W channel corresponds to the output of an omnidirectional microphone. Does that not represent pressure? Richard E (talk) 09:38, 17 May 2009 (UTC)


 * Yes, but so is the time derivative of pressure. But anyway, you're right. I did confuse pressure gradient and velocity. Sveinb (talk) 20:54, 17 May 2009 (UTC)

Further, I think that any talk of spherical harmonics in conjunction with *first-order* ambisonics is confusing rather than clarifying. Most readers will understand concepts like pressure and velocity. Not as many will understand spherical harmonics, and since first-order ambisonics can be perfectly understood without them, I think they should play a less prominent role in this paragraph than they do today.

I had my edit undone, so I will try again with a smaller edit, just correcting the factual error until we agree on the suitable role of spherical harmonics in this paragraph.

Sveinb (talk) 16:27, 16 May 2009 (UTC)


 * I have a good deal of sympathy with the view that spherical harmonics are unnecessary in describing first-order. I never used them in my entire work in the field (which is exclusively first-order), for example. Thus if the article contained only a discussion of first-order I would not feel it necessary to mention spherical harmonics at all. However, this is not the case and you need spherical harmonics to discuss higher orders. My personal preference, then, would be to describe first-order in its section without referencing spherical harmonics, and then subsequently insert a paragraph either there (or at the front of the Higher Orders section if preferred) stating that first-order can also be interpreted in terms of spherical harmonics as follows (description) and that this understanding may be useful for grasping higher orders more effectively. Richard E (talk) 09:38, 17 May 2009 (UTC)


 * As I said above, get the relationship between pressure gradient and velocity clear first (velocity lags PG), then this issue will go away. In any case, it is certainly not the W signal whose description is at fault.  I am also happy with describing first order in simple terms first, and then pointing out the relationship with spherical harmonics; but that's how it is now! Paul Hodges (talk) 10:06, 17 May 2009 (UTC)


 * Well, more or less. The pressure gradient explanation is a bit intertwined with spherical harmonics, but I wouldn't have thought to change anything if I didn't think there was an error there in the first place. Which there wasn't. Sveinb (talk) 20:54, 17 May 2009 (UTC)

Arithmetic error?
section  Relationship to coincident stereo techniques  states:


 * M = sqrt(2) * W + X
 * S = Y

However, a few sentences later the paragraph states "W = front + back + left + right + up + down (mono, omni mic)". Since M is the omnidirectional mono channel (= sound pressure) in MS-Stereo recordings, and S is the side channel, recorded with a figure-of-8 (= pressure gradient) microphone facing left-to-right (S = Y), this seems contradictory. Why should the pressure gradient X influence the pressure M in a linear combination? Maybe rather:


 * M = W
 * S = Y

? -- 141.20.195.174 (talk) 12:19, 25 November 2009 (UTC)


 * M is not usually an omni, but a forward pointing cardioid. The Mid-Side technique is described in this text, but the accompanying diagram is misleading. The Ambisonics article suggests that M is a forward pointing cardioid. HairyWombat (talk) 16:33, 25 November 2009 (UTC)


 * The term "usually" does not necessairily implies conformance to theoretical principles, right? The article you quote says "although Alan Blumlein described the usage of an omnidirectional transducer in his original patent)", which is consistent to the math operations performed by goniometers (transforming LR to MS by 45° rotation in polar coordinate space, equivalent to performing M = L + R and S = L - R). Also note the accompanying figure "Mid-Side stereo microphone technique", which clearly shows an M-channel with omnidirectional sensitivity. -- 92.225.69.109 (talk) 18:04, 26 November 2009 (UTC)


 * Not sure what your point is. Blumlein's patent is dated 1931. Modern practice is to use a forward-pointing cardioid. That is the key point. If you want to use an omni for M then, obviously, you are free to do so. W has always been defined as an omni, and has to be so because Ambisonics is isotropic. Blumlein Mid-Side (and indeed stereo) is non-isotropic. Stereo has never had theoretical rigour; recording engineers do whatever sounds good. HairyWombat (talk) 19:36, 26 November 2009 (UTC)


 * Later. Also, think about the responses after matrixing to L and R. With an omni as M, the peak responses are at &plusmn;90° (assuming centre-front is at 0°). When recording in stereo, most of the stuff you want to record is happening in front of the mics, so peak responses at the sides are unhelpful. Switching M to a forward-pointing cardioid shifts the peak responses forward, which better matches the spatial layout of what you typically record in stereo. HairyWombat (talk) 16:59, 28 November 2009 (UTC)


 * you are confusing the theoretical principle as established by Blumlein and Gerzon (Ambisonics as generalization of MS to three dimensions and higher orders) with the ad-hoc practice in daily studio engineering work. As I have to admit that this is common practice, it may make sense to mention both uses of the term in the articles. -- 92.225.71.216 (talk) 10:00, 30 November 2009 (UTC)


 * The section is titled "Relationship to coincident stereo techniques", and I do not see the benefit of including a coincident stereo technique that nobody uses. HairyWombat (talk) 17:32, 30 November 2009 (UTC)