ambisonics, quadraphonics, surround sound
I've just started reading around the subject of ambisonics, not that there is such a lot about it on the net and what I find seems quite old. I'm fascinated and would like to try some experiments but before I invest a pile of time and money into this I wondered what the diyaudio guys had to say about it. What do I find ? Nothing. Well almost nothing. What? all these people, building, experimenting with amps and tubes and DACS and speakers and tone arms and God knows what but nobody playing with surround sound. Amazing.
Is there a taboo against this? Did everyone try it already and hate it? What's up?
An impressions I get are: a) Audiophiles did not like Quadraphonics in the seventies so they gave up on the surround idea. b) Audiophiles don't much like modern cinema style surround systems so again no go. c) Surround can expensive and clumbsey. d) There is no source material.
What I learned is that a) From a theoritical point of view the Quadraphonic system can never produce a high quality spatial image. b) Modern cinema surrround is wedded to the layout of the speakers at the playback end. c) Is no issue for DIYers. d) Source material is still available form the likes of Nimbus records, can be recorded by oneself or composed with the help of a computer.
The theory behind Ambisonics is quite compelling and I'd like to put together something like:
A cubic array of inexpensive fullrange speakers, small Tangband boxes say. Yes 8 speakers for full 3-d.
Eight chip amps to drive it. I alreay have four.
A home made ambisonic decoder.
Has anyone here attempted such a thing? What experiences did you have?
I was interested once, but when I looked into the circuits that could be found, and the vague descriptions the web sites provided, I voted against doing it. Also looked at the way thye proposed layout of speakers took no concern about the impedance effects on amplifiers which was another minus point.
The 2+2+2 method from MDG seems quite easy though, which is a setup I wish to try.
Yes there is not a lot to be found on the web, I'm basically getting my info from here:
Having read through those a few time the basic ambisonic idea seems beautifully simple:
Capture the three directional components of the sound pressure at the recording position, X, Y and Z. Also capture an overall level signal W. These are recorded with a special multi-capsule mic. (or created artificially, by computer say) These signals (so called B format) contain all information about the sound field and are what is recorded, transmitted, mixed edited etc.
Importantly they relate to the moment of recording not playback. So your speaker arrangements are not dictated by your medium as in Quadraphonis systems or 5.1 Cinema sound etc.
For play back we use some simple trig to map the sound pressure signals to the available speakers. So playback can be mono, stereo, 4 speakers or six etc and of course can include height info from the z channel.
At it's simplest thats all you need and it sounds to simple. But then again normal stereo is pretty straight forward.
Then there are some frequency related compensations added because on recording the mic capsules are not truly coincident so there is an uper frequency limit on this working nicely. On playback there are phase issues as the sound runs around the listeners head. And then perhaps a correction for the fact that the maths of all this assumes plane wave fronts whereas they are of course spherical. But people seem to get by without much of this extra compensation.
The really nasty part is when you encode the B format "master" recording into stereo (UHJ format) such that it can be recorded to vinyl or CD and on playback is decoded into two dimentional surround. I don't really understand how this works yet and it involves some nasty phase shifting circuitry. BUT now a days we don't really need this, computers can easily record the 4 channels of the B format, and DVDs can carry it in their 5.1 channels.
It's that "BUT" that makes me think that now is a good time to get back into ambisonics, no more messy stereo coding/decoding anymore.
soonqsc could you elaborate on what you mean by "took no concern about the impedance effects on amplifiers"? What is the problem here?
Oh and where can I read about "The 2+2+2 method from MDG"?
Since nearly none of us are in a position to record important artists, Ambisonics will, sadly, remain a tiny niche within the tiny niche of high end audio.
This is an example of how your could wire speakers up with the ambisonics method. Seems very unlikely that you can get the same accurate results in imaging. Also the woofer signal has no reference to it's own ground, what comes out of the woofer is really dependent on what contents the other amplifiers have. It just seems to me that it uses lots of out-of-phase signals to create a synthetic sense of space, having to relevance to where the performers actually are.
Let me see if I can dig up some explanation about the 2+2+2. This site has some explanation but not much.
The basic speaker arrangement is 4 speakers in the front, 2 surround speakers. Two of the front speakers are in the normal listening location, two are higher in order to create image height. The two surround speakers are placed like normal surround speakers. The sound delay for the upper front and surround speakers are set based on distance and possibly angle from the listening location.
If you ask MDG where more information can be found about setting a 2+2+2 system up, they can tell you.
SY you are right of course, not much material. I not imagining to bin my faithfull old stereo.
But you did say you were impressed by some demos so I'd like to invest a few hours and euros into having a little of that experience. Nothing to grand.
Also I know some garage bands and other performers around here so we could have fun trying to record them.
If you're an avid diy recordist, Ambisonics is a terrific thing. You'll have fun. Four Panasonic electret capsules, some hot glue, and a few opamps will get you started cheaply.
Those speaker layouts look like the ones from the old Integrex Ambisonics kit iinstructions. Probably a good idea at the time for saving on amps. Did not like the look of it myself.
My plan would be to have four or six speakers for 2-d surround or eight for 3-d. The decoder would be built (configured) to handle the different layouts. Note the square or cubic arrangements can be stretched front to back or sideways to fit the room and the decoder adjusted accordingly.
Each speaker could be cheap Tangband with it's own chip amp. Not so expensive and no need for confusing speaker connections.
soonqsc: Your comment "It just seems to me that it uses lots of out-of-phase signals to create a synthetic sense of space, having to relevance to where the performers actually are" Seems a little harsh to me. Those out of phase signals originate from the moment of recording the original performers and are faithfully preserved (ideally) so I would not say a "synthetic" sense of space but rather a "shadow" of it's former self. Much like stereo is anyway.
Yes heater, those are from the ntegrex kit instructions. If I sound stage is to be correctly reproduced, there must be a standard was of allocating the speakers in order to preserve the time relation between the recorded direct performance and the delayed reflections and reverberation. I have not seen any information that explains how the speakers are correctly setup.
I have played around with a few phase shifting circuits that do give you a better sensation, the same feeling when you have two speakers wired out of phase from each other. If you have main speakers in phase and some surrounds out of phase or shifted some what, you will get a good sensation of space to various extent, but the important thing is that the relfection time delays cannot be correctly preserved.
When people talked about binaural recording in the early days, the mathematics was well revealed. With Amisonics, they just show some wording and some unknown matrixes that nobody could verify to be correct or not.
All this does not mean it will sound bad, but it will just not correctly preserve such experience as we are led to believe. Yes, it is the way currently stereo is.
One thing that seems interesting is that there are some software out there that allow you to create a "hall" and place each individual recorded performer in the hall and recreate the "hall effects. This is very interesting indeed and opens the door to lots of possibilities.
That integrex kit is mind bending. Seems to support a lot of different input codings that don't or never did exist and does not support the common UHJ as used by Nimbus I believe.
Not sure I understand what you say about speaker layout, direct sound and reflections. Seems to me that the basic maths of ambisonics tries to encode the 2-D or 3-D components of the sound field at the recoding position, which will then be reproduced at the listening position on playback. So for example a recorded reflection is in there just like any other sound, time delays and all. Why concern ourselves if is direct or reflected? The recorded B' format is then decoded according to whatever speaker array you have, square, hexagonal etc. Better have regular shapes, speakers equidistant from the listener.
I have to say that I have no idea about the shelf filters and distance compensation filters sometimes included to magically fix things up where the theory does not match the practical implementation. Those parameters do seem to come out of thin air! People seem to get by without them. For example this schematic for a cubic decoder:http://www.york.ac.uk/inst/mustech/3d_audio/davedec.htm
There is a lot of maths published about Ambisonics, it's accuracy etc. Just seems to be buried away in academic journals. But again the frequency dependent tweaks seem to have arrived by trial and error.
Not sure I want to get into any old phase shifting circuits for an "ambiance" effect.
|All times are GMT. The time now is 03:16 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio