Ambisonics - A New Age

Ambisonics: Overview

Ambisonics is an extension of Blumlein's original sum and difference stereo into up to three dimensions. Instead of an omni and a left-right facing figure-of-eight (L+R and L-R respectively) as in Blumlein's stereo method, you have up to three figure-eights all at right angles to each other. This gives you a mono signal (called W); front-back (X); left-right (Y); and optionally up-down (Z).

This four channel signal is called B-Format, and it is one of the things the Soundfield mic control unit produces (after a bit of processing and electronic correction for true coincidence of the mic's tetrahedral array of capsules). It contains all the information about the distance and direction of sounds picked up by the mic that you could possibly want. If you decode it on a handful of speakers, you get back a beautiful re-creation of what went on in the acoustic environment around the mic.

The Ambisonic Mastering Package (released about a year ago) does the same thing for multi track mixing: it takes mono signals such as those from a multitrack console and enables you to position them around you to create a synthesised soundfield. You can either use the console panpots - via a unit called the B-Format Converter - or you can take direct channel outs into the Pan-Rotate Unit, which has eight 360 degree panpots plus the ability to rotate the whole B-Format submix created by the panpots. And you can add in other B-Format signals and rotate them, too.

Of course, B-Format is a set of sum and difference signals, which aren't compatible with stereo and mono. It's really an archiving format, or one to use in the studio. For example, the Soundfield mic can produce a stereo signal which can be made to be any kind of stereo mic you like, pointing in any direction, and you can move the mic about electronically. If you record the B-Format signal on four tracks of a multitrack, you can still do all that stereo processing on the signal by passing it back through the control unit after you've recorded it!

So to deal with the need for compatibility, the ambisonic system includes an encode/decode system called UHJ. UHJ contains up to four channels, depending on the transmission medium available. If you have all four available, you can decode surround-sound with height on a minimum of eight speakers. With two or three channels you can decode horizontal surround (three channels - one of which can be bandwidth-limited-gives somewhat better imaging than two) on four or more speakers. With two channels, two speakers, and no decoder at all, you get a very tasty stereo - wider than the speakers and a good deal more solid, imagewise, than ordinary panpotted mixes. Sum the two channels and you have a very good mono signal - with the difference that you have all the punch you would have had if you'd mixed in mono.

For obvious reasons, there not being many multi-channel media around, we generally use two channel UHJ for most purposes. It can be cut on records, put on ordinary tape, released on CD, or broadcast with no trouble at all. With no decoder you get enhanced stereo or good mono, but with a decoder you get a very impressive surround-sound effect. As two-channel UHJ can't cope with height, we usually 'throw away' the Z channel of B-Format. The Mastering Package doesn't deal with it either. The B-Format signal from the Soundfield mic or the mixdown system is therefore carried to an encoder which turns it into two-channel UHJ. In fact, this transcoder unit is named after another of its functions: it can also 'transcode' from two stereo mixes into UHJ without going through B-Format, so you can do an ambisonie mix with no more than two stereo groups - front and rear stages, and the transcoder, which allows you to vary the width ofthe stages.

Finally, there's a decoder which derives four speaker feeds from a UHJ or B-Format input. Note that this is not a 'quadraphonic' system - you don't have to have the speakers in a square and you don't have to sit in the middle of them. You can put the speakers in any sensible rectangle and tell the decoder where they are so that it derives the right signals. In fact, you aren't conscious of speakers at all, and as they aren't at the corners of a square (they could be almost anywhere) and you aren't in the middle of them necessarily, we don't use the terms 'left front', 'right back' and so on. We think instead of North, South, East and West - and the points in between - because we're talking about where sounds are coming from, not where the speakers are (or aren't).

Back to Ambisonics - A New Age?