Talk:Ambisonic UHJ format

Verifiable content

I am concerned that, if asked to do so, we would be unable to verify this statement in the Super-stereo section:

It is also often more stable and offers superior imaging.

I know this is true (and I know you know), but can it be verified? Note that telling questioners "Go see for yourself" is classed as original research, and that is not allowed in Wikipedia articles. To see what can happen, visit 10.2 and look at the banner at the top. The Talk page includes:

This article really needs cleanup. Too much of it sounds like an advertisement. --218.234.101.44 11:36, 12 January 2007 (UTC)

I don't want this to happen to any Ambisonic pages. Martin.leese 23:33, 14 January 2007 (UTC)

Valid point. Presumably we need to cite somebody. I didn't invent UHJ so if I've said that in any of my articles you could cite me. The JAES is a peer-reviewed journal so if anyone has said that in there (including the inventors) then that would do. If there was anything in the patents that made that claim then one could presumably cite that if the patent was granted. Meanwhile I'm going to delete this statement here for the time being.

In fact there is a lot more citing of articles we could do here and in the original piece. I do not have most of the original articles available any more: we really ought to go through your publications list (all of which could presumably be used as references - perhaps the list should be run as a separate article) and tag entries in the articles appropriately. Richard E 10:09, 15 January 2007 (UTC)

While we could go through and cite everything, it would be a lot of work. A better approach, I would suggest, is to identify those statements which might be questioned. Apart from the one here (which you have removed for the moment) the only other place where I can see trouble is the Advatages and Disadvantages in the main article. I kept this is mind when I re-wrote them, but some might still cause trouble. Martin.leese 21:28, 15 January 2007 (UTC)

Matrix HJ

Martin, in the Quad-Matrix H section you (presumably) state: (Matrix HJ) "was based on zones designed to include both Matrix H and the two-channel UHJ encoding of Ambisonics." I have never heard of this and I wonder where it comes from. I also don't see how it would be possible in a 2-ch matrix, but I am sure you can provide a reference. Ta! Richard E 14:06, 17 January 2007 (UTC)

The Matrix H section already existed, but did not mention Matrix HJ. The reference is a Wireless World article published back when the BBC released details of Matrix HJ. (Unfortunately my copy is still in storage at the other end of Canada.) As I understand things, Matrix H and Ambisonics were completely separate, The BBC were then criticised for not collaborating with the Ambisonics team. Under pressure they did, failed utterly to reach agreement, and fudged a new standard called Matrix HJ. Matrix HJ was not a single encoding. Instead it allowed a range of possible encodings, the range (zone) being sufficiently wide enough to allow the BBC to continue doing exactly what they had been doing before. Martin.leese 20:54, 17 January 2007 (UTC)

Later. I should probably explain that when I started writing the Ambisonics FAQ in 1994, I asked the library at the university I was working for at the time to drag up from their basement about twenty years worth of Wireless Worlds. They filled an entire library cart. I then spent an afternoon looking at each copy, and photocopying everything to do with quadraphonics or Ambisonics. Therefore, my recollection of the Matrix HJ zones is from only 10 years ago, not 30 years ago. Martin.leese 21:29, 17 January 2007 (UTC)

Great. My knowledge of the lit definitely is 30 years old - and this bit I had never heard of before! So thanks for increasing my knowledge in quite a significant way! Richard E 10:38, 18 January 2007 (UTC)

I remember the report in Wireless World with the diagram showing the zones from 30 years ago, and indeed still have my copy. I also found this quoted reference: "BBC Engineering information department, Publicity section: Additional information on BBC/NRDC System HJ, System HJ tolerance zones." on the Ambisonics Publications page of the Ambisonics UHJ Discography, but I don't know if it is still possible to get hold of this. Paul Hodges 16:41, 29 January 2007 (UTC)

Great. Please read what I wrote over on Quadraphonics#Matrix H and see if what I wrote from memory is good. In particular, please check to see if the Matrix HJ zones included UHJ or its progenitor System 45J. Martin.leese 00:15, 30 January 2007 (UTC)

Reading what I wrote, while what I wrote is factually correct, I was probably unduely cynical about the BBC's motives. The main concern of the BBC was that existing mono and stereo listeners should not suffer a degradation in sound quality when listening to a quad broadcast. I am sure that they didn't just want to continue doing what that had been doing before.—Preceding unsigned comment added by Martin.leese (talk • contribs) 17:19, 18 January 2007 (UTC)

Decoding Equations

Richard Lee pointed out on the Sursound list (5 May 2011) that the decoding equations presented here had a spurious factor of root 2, and so didn't match the reference (Gerzon 1985) from which they were supposedly taken. I have corrected them to match the reference, but have still left out the extra complexities that Gerzon includes in the three-channel case - because in the absence of any such material it seems pointless to present them. I have left the encoding equations alone - I presume that someone had good authority for the greater precision in the coefficients compared with Gerzon's. Paul Hodges (talk) 21:07, 5 May 2011 (UTC)

The greater precision was me, taken from a Gerzon paper. Unfortunately I no longer recall which one. As it stands, there is still a mismatch with the Gerzon 1985 paper in that, in the paper, three-channel UHJ uses W', etc, not W (I don't see this as a problem). Also note that, as currently presented, the four- and three-channel UHJ equations are the same (apart from the fourth channel), and so could be usefully combined. HairyWombat 14:02, 6 May 2011 (UTC)

It just seemed to read more simply having three and four channels separate; it also leaves the option for adding the extra refinements involving t, B' and the k factors. I left the three-channel case as W, X, Y for the time being, rather than W', X', Y', because in the form given the values are exact (i.e. with t=1 and k=0). Mind you, since there is AFAIK no such material in existence, there is even an argument for reducing the article to two-channel only, mentioning the three and four channel cases as theoretical matters that were never taken further. Paul Hodges (talk) 22:30, 6 May 2011 (UTC)

I strongly agree that the variable t should be fixed at 1. The trend these days is to not transmit analogue signals, replacing them with digital, so the probability of an analogue T-channel ever being broadcast over the airwaves is zero. There is still an argument for the use of three- and four-channel UHJ, as they elegantly combine stereo and surround sound. However, there are problems as, at the moment, a stereo system that doesn't know any better will add the T and Q channels into its stereo mix instead of simply discarding them. (There is a fix for this, but so far nobody has been interested in promoting it.) HairyWombat 00:30, 7 May 2011 (UTC)

I believe the long-term solution is to move all the decoding stuff over to the Ambisonic decoding article. It belongs there anyway (along with the intricacies of B', etc). Before that can be done, however, the Ambisonic decoding article will need to be reorganised. At the moment, I don't see how to do that. (I had always hoped that somebody who designs decoders would re-write the Ambisonic decoding article, but alas not.) HairyWombat 14:02, 6 May 2011 (UTC)

I am neither a mathematician nor a decoder designer - my contributions are entirely based on my reading skills! Paul Hodges (talk) 22:30, 6 May 2011 (UTC)