2 speakers to play binaural recordings or even surround sound!
Ambiodipole / Stereo Dipole / Transaural
- Richard Lee 15dec07

This page is being revised. It is presently a personal view of how things transpired. But please visit links at the end for more info and alternative viewpoints and histories. Many thanks to Jerry Bauck & Ralph Glasgal for much background and history which I have yet to incorporate.

In particular, it is now obvious that Bauck & Cooper were probably the first to propose and demonstrate X'talk cancelling in the form which gives substantial advantages over normal stereo or binaural.

Various people's comments are from the Sursound forum circa 2005 - 2006.


Ambisonics is surround encoding which makes no reference to speakers. Instead Ambisonics encodes the soundfield, ie what you should hear, including the vertical information. What you feed the speakers depends on how many you have and where they are. is [was - not currently available] a treasure trove of true surround sound recordings of very high quality. Includes DTS surround recordings you can play on a HT computer or burn to a CD and play on your HT system.

Permanent Ambisonic Systems maintained by members of sursound; the Surround Sound forum which has been running since 1995.

I'm sure most, if not all of you, are experienced at using headphones to monitor when recording live. Even when you have the place to set up a pair of speakers in stereo (let alone surround), invariably, you can't use it while the recorder is running cos it is too close to the actual venue and can only use it for the musicians to hear the last take.

You must have been impressed by the sound from a good pair of headphones and wondered how it might be possible to achieve this with speakers. eg I LIKE fig-of-8s on headphones even with the distortions mentioned by Erdo Groot.

These distortions disappear if you use a dummy head ie make a binaural recording which captures everything the Mk1 HH (Human Head) would hear and sends it straight to the punter's pinnae.

This is perhaps the 'purest' form of record/playback with the least theoretical faults. And it is a SURROUND / Periphonic technique though how this is done is complex.

But binaural has one very important shortcoming. It is a fixed head technique. It doesn't capture the important moving head cues.

Playing this back on speakers requires that each channel reaches the correct ear and NOT THE OTHER EAR.

In the early 90s, DSP became powerful and affordable enough to be worth investigating for these X'talk cancellers. This was first tried for the usual +/- 30° stereo speakers by Schreoder using an IBM mainframe to pre-process the signals.

Kirkeby, Nelson & Hamada of the ISVR did a lot of work on X'talk cancelling DSP filters used to prevent the signals from reaching the wrong ear.

This gave impressive results but the sweet spot, both angular & spatial was VERY limited; much worse than normal stereo. Exactly what was expected and suggested by Dave Hunt.

But the breakthrough was when X'talk cancelling was tried with 2 speakers close together. To everyone's surprise, the sweet spot became much larger. It is a lot better than normal stereo and comparable to specials designed specifically for greater sweet spot like the Wharfedale Option 1

Angelo reports

The good property of a Stereo Dipole system (in which the loudspeakers of each pair are closely spaced, something as +/- 10°) is its robustness to head movement and rotation.

In a good Stereo Dipole system, the cross-talk cancellation remains very effective for movement back and forth of approximately 1m, up and down of approximatley 0.5m, left-right of 3-5 cm, and rotation of +/- 30°.

The ISVR call this the Stereo Dipole

When you turn your head, the sound stage DOESN'T MOVE WITH YOU. Somehow, by serendipity, using the Stereo Dipole to play binaural recordings has given us a reasonable facsimile of natural moving head cues.

But you can't turn and face the sides as you can with an Ambisonic system .. and height is a problem.

Double Stereo Dipoles (ISVR 4x4) are an extension of this. It can increase the angular sweet spot to +/- 50° but reduces the back/front sweet spot.

As is the ISVR Optimal Source Distribution (OSD) to reduce the demands made on the speakers at LF. The OSD shows pleasing similarities with 1st order Ambisonics at the appropriate frequencies.

So here we have an efficient (2 channels only) Surround coding system which incorporates pinnae & ITD cues in a holistic and theoretically satisfying manner. Much more satisfying than ORTF or the offerings of other pseudo prophets ...

Complicated pictures & (fairly) non poly-syllabic explanations on the above ISVR web site.

Alternative Viewpoints & Histories

Jerry Bauck and the late Duanne Cooper, one of the co-inventors of Ambisonics, used analogue circuits for X'talk cancelling in the 80's.

They were the first to propose (1987) and demonstrate the "speaker close together" solution with DSP x'talk cancelling. ie the Stereo Dipole. You need to be a member of sursound to see this page.

Also an extended source for LF power handling (AES jun97 Seattle) which is the basis of the OSD. The ISVR usually reference their work but this is often hidden in the fine print!

Their long collaboration includes Dr Bauck's development of virtual home theatre (analogue) for Harman Int.

Ralph Glasgal experimented with acoustic means of X'talk cancelling long before the dawn of the digital age and Ole Kirkeby was surprised to find he already knew the benefits of keeping the speakers close together.

Today, with Angelo, he prefers to use Dave Wareing's simplified HRTF, which is for a sphere and two holes on the side without pinnae, for his X'talk cancellers and reports less problems than more complicated dummy heads. What he calls the Wareing Ambiodipole may be the most robust implementation of the Stereo Dipole.

Very impressive demos. The Ambiodipole system has been extended to include height.

Ralph points out that Don Keele & TM Bock may have been playing with close speakers outdoors and acoustic barriers at about the same time.

AES preprints 2420 A & B nov86

David Wareing seems to have disappeared from the known universe. Professor Edgar Choueiri, Applied Physics Group, MAE Dept. Princeton University has taken up the banner of advanced Crosstalk Cancellation (XTC) and is presently (dec07) preparing a paper on Optimal Crosstalk Cancellation for Binaural Audio from Two Loudspeakers which he has dubbed BACCH.

Robert Greene, who writes for The Absolute Sound, is likely the first to put the idea of X'talk cancelling with an acoustic barrier in print.

Ambisonic Info header