Michael was keen to explore the possibilities of multi-channel recording. In particular, he wanted to record faithfully the ambience of the live performance, by capturing the sound arriving from all directions (including from above and below) at the listener. This would help to overcome a weakness of ordinary two-channel stereo, in which all the direct and indirect sounds appear in a narrow horizontal sound stage in front of the listener. This is especially so for the Blumlein microphone technique favoured by Michael, in which sounds from the sides and behind the listener are captured by the figure-of-eight microphones but reproduced from in front (see Microphones and scroll down to Calrec Microphones sub-heading). The four-speaker system described in the Surround Sound section was remarkably successful in reproducing sounds from the sides and rear which had been captured using the Blumlein microphone technique and recorded on only two channels. Michael’s analysis described below demonstrated that an increase in the number of channels would allow more comprehensive and accurate directional information to be captured.
Principles of Quadraphonic Recording
Commercial quadraphonic systems assumed that four channels were needed in order to convey horizontal surround sound to four loudspeakers, but Michael showed that this assumption was incorrect. He published this finding in the first of two articles entitled ‘The Principles of Quadraphonic Recording’ which appeared in Studio Sound in August 1970 (Vol. 12, pp. 338-342). [Principles 1] The subtitle of the first article asks the question ‘Are Four Channels Really Necessary?’ Michael’s answer was no, they are not, and showed that only three channels are needed.
Whenever a coincident microphone technique is used for sound to be reproduced over four loudspeakers arranged horizontally around the listener, only three microphones are actually needed to obtain all the audio information. The sound fed to each of the four speakers can be derived by suitable matrixing of the three microphone signals. Thus, for many purposes, only three recorded channels are needed to convey all the information reproduced by the four loudspeakers.
He concluded that since commercial pressures made it likely that practical quadraphonic media would in fact convey four channels, other information might be ‘smuggled’ into the system. Michael’s proposal was that ‘conventional (!) quadraphonic recordings can be used to reproduce height information via suitable reproducing equipment.’
The second article was subtitled ‘The Vertical Element’ (Studio Sound, September 1970, Vol. 12, pp.380-384) [Principles 2]. Michael extended his arguments to the reproduction of sound from all spatial directions about the listener, both horizontally and vertically. He christened such reproduction techniques ‘periphonic’. First, Michael had to counter the argument that height information was unnecessary. Some believed that the human ear was insensitive to height information, but Michael quoted research which showed that ‘it is possible to perceive the elevation of a sound quite accurately by means of small unconscious head movements’. He went on to state his belief that ‘height information can be of great musical importance’, and gave a number of examples including orchestral and choral music, and organ music where the instrument is high above the listener.
Michael was also able to dispose of the argument that it was undesirable and impractical to introduce more than one system of quadraphony, by showing that the periphonic recording was ‘capable of being reproduced via a conventional quadraphonic set-up’. He continued by explaining that no matter how many loudspeakers were used to reproduce the sound, only four microphones were needed to pick up all the periphonic audio information that could be obtained from coincident microphones. He went on
The requirements of simplicity and compatibility with conventional quadraphony have thus led to a periphonic recording system in which four identical hypercardioid microphones point along four tetrahedral axes.
After considering the required microphone layout and directional characteristics in detail, he concluded that
The best results would…probably be obtained by using four microphone capsules placed in close proximity, and a tetrahedral arrangement of four hypercardioid capsules placed back-to-back should prove satisfactory.
In conclusion, he stated that
It has been shown that it is possible to convey periphonic sound via (four) channels … that can be reproduced via the horizontal quadraphonic ‘box’ speaker layout as in current American proposals, or via a tetrahedral loudspeaker layout giving three-dimensional sound reproduction over an exceptionally large listening area.
Finally, he made a plea that
it would be wise for recording organizations to include height information on current quadraphonic master-tapes, to allow for the possibility that periphonic systems may become commercial.
Experiment Tetrahedral Recording
Michael concluded his second article by emphasising that ‘the work is mainly the result of a theoretical analysis’. He immediately set about organizing a practical experiment to test these proposals. We had only just started experimenting with the three-Calrec microphone set-up (see Microphones and scroll down to Calrec Microphones sub-heading) when he ordered a fourth Calrec to allow a tetrahedral arrangement of microphones for four-channel periphonic recording. He also ordered two Spendor BC1 loudspeakers to use as the raised pair in the tetrahedral playback arrangement, because he felt that it would be impractical to suspend Quad Electrostatic speakers about 2m from the floor! Unfortunately, purchasing a four-channel tape recorder was out of the question both for OUTRS and Michael personally as even the cheapest machine was unaffordable. This was a major stumbling block since recording and playback experiments were forced to depend on borrowed equipment.
He planned to set up a live relay and recording of a concert given early in May 1971 [8th May] by the Schola Cantorum of Oxford in Merton College Chapel. The venue and performers were well known to us which was ideal for comparison with previous sessions. But by early April he had still not been able to arrange the use of a four-channel tape recorder, and there was uncertainty until just before the concert.
Meanwhile, Peter agreed to design an individual amplifier and matrix circuit to convert the output of the microphones from cardioid to hypercardioid, together with a circuit for correcting the peak in the treble (an equalizer), for each of the four Calrecs for tetrahedral use. We called these circuits ‘monocronics’ individually, and collectively the ‘Quadracronic’, since four of them were built into one box (see Microphones and scroll down to Calrec Microphones sub-heading for explanation of terminology). He also designed differential amplifiers for producing a Blumlein figure-of-eight pair of outputs for stereo recording from the output of the four cardioid Calrecs.
Most of this equipment was constructed by Richard Cowderoy, who had rapidly become an indispensable member of the team since arriving as an undergraduate the previous autumn.
A week before the tetrahedral session, the Quadracronic and differential amplifier were still not finished – Philip, who lived near London and was visiting Oxford for the weekend, was pressed into buying components and delivering them to us. Every spare minute was spent drilling holes in boxes, soldering components and testing the result.
The day before the concert, we recorded an organ recital in Merton College Chapel (in two channels only), and to our amazement, the Quadracronic worked straight away! Unfortunately, the differential amplifiers were finished too late to obtain figure-of-eight outputs, so we had to record using just two Calrecs in cardioid configuration.
The music, performed by an orchestra and the Schola Cantorum of Oxford, conducted by Andrew Parrott, included choral works by Buxtehude, Gabrieli, Bach and Monteverdi. The highlight was Mozart’s Coronation Mass (K.317), a choral and orchestral piece that the choir performed again a few weeks later for a commercial recording with the Academy of St Martin-in-the-Fields conducted by Neville Marriner (issued on Argo ZRG 677).
The recording and playback equipment was set up in the Sacristy, adjacent to the Chapel. We could easily pass through a door into the Chapel for comparison of the live sound and its reproduction via loudspeakers.
The live relay and recording of the concert was remarkably successful, considering the unusual complexity of the technical arrangements, the last-minute completion of equipment, and the many guests.
Everything appeared to go smoothly, though Michael was annoyed because of a fault on the Crown [tape] machine. A half-inch Scully [tape recorder] appeared from nowhere in the afternoon, made a tape, and as mysteriously disappeared. Michael bought the tapes and went off to gloat over them! Rex Baldock came to do a piece for Hi-Fi News, and Roger Driscoll for Wireless World. Granville Cooper and Sid O’Connell came from Guildford. [Cooper had carried out tetrahedral recording experiments at the University of Surrey.] ‘Baldock is a very nice chap who’s at least prepared to meet us half way on some of our views e.g. on loudspeakers and microphone technique – and he doesn’t like the Dolby!
But it was impossible to find common ground with Driscoll – ‘even Michael, the Arch-Arguer, eventually realised how futile it was.’
We started with a hilarious effort to set up a proper tetrahedron [for the loudspeaker arrangement] – it kept coming out as a rhombus, and we had to use a long piece of rope to measure out the sides. We used stepladders to get the Spendor [speakers] to the required height, and when we tested it all out we had Michael appearing from the rear right Spendor [above our heads] and I said “thus spake the voice of God from on high”, which amused Rex Baldock no end!
During the rehearsal,
we all alternated between the Sacristy and the Chapel to compare the sound. I was quite pleasantly surprised - [the sound] was more spacious than I’d expected. There was trouble at first from a sloping image, which was improved by careful adjustment of volumes and balances. The sound was also improved later by moving the mikes.
The live sound was
over-bright and harsh at first, though better at the performance due to the audience. We moved the mikes further back for that [the performance]. [On the relay] there was more volume available than I expected without gross overload, even on input, though quite a lot of nastiness from Quad amps. The Spendor [speakers] gave out a lot of colouration, more noticeable on sounds coming exclusively from them, but quite pleasing overall sound on choir. Also sounded pleasantly spacious outside the listening area too. But still had impression that sound stopped at the loudspeakers – we obviously need an eight Quad [speaker] matrix for playback! The performances seemed good, with the choir as a whole in good form.
After the performance, there was a
massive clearing–up operation – the visitors kept talking [to Michael] while Richard and I and other OUTRS people, including Philip [visiting for the weekend], helped to move all the stuff – it took till 12.30 am.
The following day we set up a playback session in a student’s room in Merton College so we could hear the four-channel recording before the Crown tape machine was taken away. Peter decided that he wanted to use an all-Quad speaker playback arrangement instead of ‘nasty Spendors’, so rushed back home to fetch his own pair. These were strapped to the top of step-ladders, with Michael groaning in alarm! Richard and a friend were ‘very energetic, moving furniture and speakers with lightning speed, while aged and decrepit Peter, Michael and I stood and watched!’ It took so long to set up that there wasn’t time to play much music, but the Quad speakers were felt to be an improvement over the Spendors. The lack of control over the way the four-channel tapes had been recorded was a problem – Michael was not happy with the way the 6.25mm tape had been made, and he tried to get a 6.25mm copy of the 12.5mm master tape.
A few days later we received a draft of Rex Baldock’s article for Hi-Fi News. It was disappointing –
too much waffle on “great piles of gear” and “the musicians enjoyed hearing the applause coming from all round”. We feel it should include a statement of the aims of the experiment and his views on the achievement. He was very non-committal on that, just saying it was “a pleasant spacious sound”, then criticising faults like “slope” and “inadequate volume”, while privately he said it was the best four-channel he’d ever heard.
Michael was equally quick to prepare the first of three articles which were published in Studio Sound, entitled ‘Experimental Tetrahedral Recording’. He showed us the draft, which I thought (rather patronisingly, with hindsight) ‘quite a good article – brings out the main points well’. Part 1 appeared in August 1971 (Vol. 13, pp 396-398) [Tetrahedral 1], and concentrated on ‘a description of the experimental set-up and an account of some impressions obtained by listeners.’
Michael focussed on ‘two really important flaws’ that were heard. ‘The first defect, given the name “overlap” by Rex Baldock at the time, is the effect obtained when the sound corresponding to one direction emerges to some degree even from speakers in the opposite direction.’ He showed that overlap was due to a sound arriving from the front at the cardioid microphones being picked up by the rear microphones only 11.4 dB down relative to the front microphones. In the experiment, hypercardioid microphones were simulated electronically, and it was found that this gave a considerable reduction of overlap.
The other important flaw in the tetrahedral reproduction is far more difficult to rectify. Even when the levels were set optimally, it was found that the colouration from the four speakers caused a very disturbing side-effect. Although the basic stereo image was distributed horizontally, the four loudspeakers were heard as separate and very disturbing sources of colouration…These sources of colouration greatly disturbed the overall impression of a homogeneous sound field around the listener, and this is certainly the most serious problem to be solved with this system.
He even suggested that tetrahedral reproduction is an ideal way to test loudspeaker colouration – ‘there is clearly an enormous amount of progress yet to be made in loudspeaker design.’
Michael went on to describe the subsequent playback session in which four Quad speakers were used.
This set-up gave much better results… the speaker colouration was found to be less disturbing, although its effects were still noticeable… It was found that, with the all-Quad system in the domestic room, the original acoustics of the chapel were audible with great clarity…It was possible to analyse the acoustics in the same detail as if one were there live…the whole experience strongly argued against those who claim that four channels need only pick up a generalised reverberant richness and nothing more.
He noted that
the height effect on the reverberation added very considerably to the realism. Indeed, several listeners standing outside the tetrahedron still found the spaciousness of the recording to be superior to that obtained from most conventional four-channel recordings within the square of speakers
Michael pointed out that small performance errors were less noticeable on the tetrahedral system than on two-speaker stereo playback. Characteristically, he emphasised that the ‘purely musical value of tetrahedral reproduction should not be underestimated’
These have been some initial reactions to skew-tetrahedral reproduction. It offers a tantalising glimpse into what audio could be like, and one becomes depressingly aware of the overwhelming deficiencies of even the best conventional four-channel stereo. It, or something like it, is clearly the system of the future, but how far in the future is anyone’s guess…
During a subsequent discussion with OUTRS members, he was asked whether tetrahedral reproduction might become a domestic proposition. He replied that although horizontal surround sound was achievable, he thought that with-height reproduction
might take 10 years because loudspeakers aren’t perfect. Colouration is so objectionable that the system would be unusable for domestic use – bearing in mind that the Quad is one of the least coloured speakers available.
Tantalisingly, he added that he had ideas for how to overcome these problems… (Quote taken from an OUTRS talk and discussion on C236/122).
The second part of the series appeared the following month and stepped back to consider the problems facing anyone trying out similar experiments, and their solutions (Studio Sound, September 1971, pp 472, 473 and 475) [Tetrahedral 2]. Michael first considered the type of tetrahedral loudspeaker layout that would be used for playback. After describing various alternatives, he stated a preference for the skew tetrahedral layout used in his experiment, because all its speakers lie at the same angle off the ear’s axis, it has a large volume for a given room height, it is less liable to a hole-in-the-middle effect, and provides more realistic information to human stereo location mechanisms. He felt these advantages outweighed the unsuitability of the arrangement for reproducing two-channel stereo. However, he admitted that ‘the one big disadvantage of the skew tetrahedron system is that speaker colourations emerge from directions quite different from those associated with direct sounds…’, and gave some suggestions for experiments to overcome the colouration problem.
Michael dealt with ‘the tricky problem of microphone technique…a profound philosophical problem with tetrahedral recording is where to put the microphones.’ Because the aim of the system is to reproduce the live sound, he advocated ‘placing the microphones at a sensible listening height’, rather than suspended several metres in the air. Michael also showed how a differential amplifier can be used to derive a Blumlein (i.e. 90 degree angled crossed figure-of-eight) signal for simultaneous two-channel recording. Finally, he stressed the importance of not varying the gain of the rear channels independently of the front, and of not applying dynamic compression to the recording, in order to allow a true comparison with the live sound.
The final part of the series was published in Studio Sound in October 1971 (pp 510, 511, 513 and 515) [Tetrahedral 3]. Michael described the possible uses of tetrahedral recordings in a wide variety of playback experiments. He pointed out that if the recording was made as described in Part Two of the series, i.e. consisting of four coincident cardioid or hypercardioid signals pointing to the four corners of a cube, it is possible by matrixing these four signals to obtain any conventional microphone characteristic output pointing in any direction. He went on to describe a matrix circuit by which it is possible to derive any combination of the four input signals, and most of the rest of the article is devoted to giving the matrixings required to derive various different types of signals from the standard tetrahedral recording. Tables of matrix values are provided in order to allow playback via one of the other tetrahedral layouts, and via arrangements of six or eight speakers. Michael noted that the possibility of such loudspeaker layouts meant that the name ‘tetrahedral stereo’ was a misnomer, and reiterated his proposal ‘that systems of recording the full directional effect around the microphones, including height, should be called periphonic systems.’
Michael pointed out that another use of tetrahedral recordings was in the study of two channel stereo microphone techniques.
A tetrahedral recording made with coincident microphones contains within its four tracks sufficient information for any conventional coincident microphone recording to be reconstructed by matrixing. Thus, for the first time, it is possible to perform repeatable objective comparisons between the different microphone techniques.
The above account has only indicated a few of the many possible experimental uses of tetrahedral recordings, but nevertheless indicates just how much information is contained in the four channels. In a precisely definable sense, tetrahedral recording makes much more efficient use of four channels than any other current proposal. So great is the system’s flexibility that a full appreciation of its uses and possibilities requires a more profound analysis than is possible in these pages. This flexibility is equally great whether coincident or multimike techniques are used. A great deal of experimental work remains to be done before the system is ready for domestic use, as is apparent from the very large number of possible playback methods.
The tetrahedral recording session had a profound influence on Michael’s ideas on periphony, and led directly to his subsequent work on Ambisonics.
In the meantime, Michael turned his attention to ‘the overwhelming deficiencies’ of commercial quadraphonics.
Michael Gerzon, ‘The Principles of Quadraphonic Recording - Are Four Channels Really Necessary?’ Studio Sound, Vol. 12, pp. 338-342 (August 1970) [Principles 1]
Michael Gerzon, ‘The Principles of Quadraphonic Recording - The Vertical Element’, Studio Sound, Vol. 12, pp.380-384 (September 1970) [Principles 2]
Michael Gerzon, ‘Experimental Tetrahedral Recording: part one ’, Studio Sound, Vol. 13, pp 396-398 (August 1971) [Tetrahedral 1]
Michael Gerzon, ‘Experimental Tetrahedral Recording: part two ’, Studio Sound, Vol. 13, pp 472, 473 and 475 (September 1971) [Tetrahedral 2]
Michael Gerzon, ‘Experimental Tetrahedral Recording: part three ’, Studio Sound, Vol. 13, pp 510, 511, 513 and 515 (October 1971) [Tetrahedral 3]