Stereo Society - albums, artists, archives - page banner

Recording Gurrelieder In Surround Sound

By Paul Myers (Director, CBS International Masterworks) and Bob Auger (Bob Auger Associates)
originally published in Studio Sound (now defunct), June 1975

In fall 1974, in one of the most complex sessions London has hosted, Arnold Schoenberg’s Gurrelieder was recorded by CBS for stereo and eventual quadraphonic surround release. The musical, production and engineering background is covered, from both stereo and quadraphonic viewpoints.

Session photograph, Pierre Boulez conducting the BBC Symphony Orchestra and Chorus at the Methodist Central Mission in West Ham, London, in 1974:

Pierre Boulez conducting the BBC Symphony Orchestra and Chorus at the Methodist Central Mission in West Ham, London, in 1974

Part one: Production
We were listening to a final playback of the two-track mix and, as the orchestra swelled to yet another huge climax (and our eyes scanned the vu meters nervously), Bob Auger smiled and said: ‘Well, it certainly sounds expensive’. I suppose this was a slightly less than respectful comment, but I am sure you will understand our reaction. There seemed nothing more to add at the end of a long and tremendously difficult project -– at times frustrating, at times awe-inspiring, but always exhilarating. And we had reached the penultimate stage: all that remained was to see whether the cutting engineers could transfer the information faithfully to a disc, despite sides lasting 29 and 30 minutes.

Schoenberg’s Gurrelieder, a work just short of two hours and scored for six soloists, a gigantic orchestra and a battery of mixed choirs, must be one of the most challenging projects for any producer or engineer. Because of the size of the forces involved, it is seldom performed (although London has enjoyed two Promenade Concert performances, the last of which was the highlight of the 1974 summer season). However, perhaps because of these concerts and certainly in honor of the 100th anniversary of the composer’s birth, a previously empty catalogue has seen the appearance of three recordings of Gurrelieder: a recent EMI issue, a reissue from Deutsche Grammophon in their Privilege series, and the new CBS production. The other recordings are radio broadcasts, transcribed to discs (although I understand the DG version was compiled from several rehearsal and broadcast tapes). In the case of the CBS recording, the decision was taken about two years ago to make the first studio recording of the work, to appear in both stereo and quadraphonic ‘surround’ versions.

At the risk of offering superfluous information, let me add a few words of musical background to the piece. Schoenberg originally composed a set of songs after poems by the Danish botanist/poet/novelist Jacobsen and entered them in a competition in 1899. He was very short of money and, despite the fact that the songs did not win the hoped-for prize money, the composer decided to expand upon the original material. He set the entire Jacobsen cycle, re-scored if for the enormous forces mentioned, and added a few more sections of his own. The work was composed by 1901, but he only worked sporadically on the scoring, which was not completed until 1911. The première was given in Vienna in 1913, to a very enthusiastic audience and, although he was present, Schoenberg apparently took no great pleasure in the reception. He was already composing in his ‘new’ serial style, and receiving little praise or encouragement. Gurrelieder looked back to his earlier, discarded Romantic style. Its Wagnerian echoes, harmonically and even melodically, its use of leitmotivs and its great washes of sumptuous sound are almost diametrically opposed to the music that followed.

Gurrelieder is, perhaps, one of the last great Romantic masterpieces (and anyone who examines the score, perhaps adding the extraordinary analysis of the work by Alban Berg, will agree that it is a veritable masterpiece by a 26-27 year-old composer), but I have always suspected that, the huge forces notwithstanding, it would have become far more famous and far more frequently performed had it been composed by any other man. For Schoenberg’s later music has long struck terror in the hearts of average concertgoers, whose tastes fun (and why not?) to more easily distinguishable melodies and readily familiar harmonies. His music may inspire and delight professional musicians, but its austere dissonances dismay audiences and empty box-offices. Only Verklärte Nacht,another early work from the same period as Gurrelieder, has attracted listeners with its instantaneous beauty, and there are one or two other, lesser-known pieces. But most Schoenberg, early or not, is generally avoided in case it proves to be similar to the last, atonal works: craggy, unrelenting pieces like the opera Moses And Aaron. I am not suggesting either that, if you like early Schoenberg, you will every like late Schoenberg, but hope that, when confronted by an unfamiliar work by him, listeners will be tempted to check the date of the composition or see whether it bears a low opus number.

To return to the recording, perhaps the best way to sum up the major problem is to show what sort of an orchestra we used. The BBC Symphony Orchestra, suitably augmented, included 24 first violins, 20 second violins, 16 violas, 12 cellos, ten basses and four harps. In other words, the strong section alone was as large as the average London orchestra. To this were added 25 woodwinds, 12 horns, 15 brass, two sets of timpani and a percussion section requiring ten players. The mixed choirs numbered well over 300, and the total number of musicians involved came to just under 500. The figures are not produced to dazzle, but point to a basic recording problem. Readers of this magazine understand the problems of reproducing the dynamic range of a normal orchestra for records, and the consequent compression of that ‘real’ range to reproduce a ‘natural’ recorded range, so I will not waste space with long explanations and diagrams. Suffice to say that I like to compare recording techniques with photography. An 80-piece orchestra can be reproduced on a record (or a tape) with the same degree of accuracy and fidelity to its ‘natural’ sound as a 2 1/2 x 3 1/2 in snapshot of the Grand Canyon reproduces a 300-mile hole in the ground. We accept that each suggests what it reproduces rather than actually reproduces it. The Gurrelieder makes use of an orchestra almost twice the size of a normal symphony orchestra (and the massed choirs), but there are many times when the singer is accompanied by little more than a string quartet of first chair players, or a small section of the orchestra. Somehow, the recording must convey (at a reasonable level) both the effect of the small group and of the full group, without asking the listener to sit next to his amplifier, suitably raising and lowering the volume controls. (Only recently, I read a review – of one of the other recordings – in which the critic made this particular comment, suggesting that this was the only way one could really hear the work. I hope we have persuaded him otherwise.)

The next problem was to find a hall that could accommodate an orchestra and a choir of this size and still lend itself to recording with some sort of ambient acoustic. Most London halls, filled with these forces, would become totally ‘dry’ and unflattering. A natural choice was the Albert Hall (where CBS recorded the Verdi Requiem with Leonard Bernstein and the London Symphony Orchestra) but it was unavailable for the period, and we spent some harrowing weeks searching for anything from a cathedral to an aircraft hangar. Finally, Bob Auger came up with his own ‘discovery” the Methodist Central Mission in West Ham, a large and superb hall, very similar in design to Kingsway Hall (but cleaner), with a circular auditorium, a domed ceiling, and a gallery that could accommodate up to 500 singers. The hall proved to be one of the best I have ever worked in, with a glowing combination of great clarity and airiness, and I am quite sure that, like all fine recording halls, it would be just as excellent for recording a solo piano or a string quartet. Even with the forces we used, there was additional space, giving us the clear, open sound we wanted.

Faced by that frightening number of players, I elected to record the work on 16 tracks. I know that engineers will argue from now until the end of recording the relative merits of multiple-track recording (and, when all is said and done, there was a time when we had to put it all on one mono track), but I favor making use of whatever new technology is available. As long as the final balance is satisfactory, it really makes no difference whether a classical recording is made on two tracks or 24, and I believe that the use of the Dolby system (on a well maintained desk) will achieve sound of the highest possible quality. Multiple-track recording permits one a second chance for the ideal balance during the remixing sessions, especially as there is never enough sessions time during a recording session for the ideal number of musical ‘takes’ and balance rehearsals. With the support of a 16 track tape, and the possibilities of delicate rebalancing where necessary, a producer can undertake a work of this size and stature with an extra degree of confidence. (Perhaps I should add, as a happy afterthought, that when it came to remixing the 16 tracks, there were very few occasions when we had substantially to alter our original settings. The 16 slider settings on the Neve console made a straight line, which is no small tribute to the excellently modulated and controlled tapes that Bob Auger created during the sessions.) The further advantage of 16 tracks was that they enabled us to split the very large woodwind section into two tracks (needless to say, the strings had one track per section), instead of attempting to balance all of a physically widespread section on a single track.

The track layout (selected by Bob, who also undertook the remix sessions) was as follows:
1. Horns
2. Flutes/clarinets
3. Oboes/bassoons
4. Brass
5. Percussion
6. Timpani
7. Harps/celeste
8. Solo Voice
9. 1st Violins
10. 2nd Violins
11. Violas
12. Cellos
13. Basses
14. Chorus
15. Chorus
16. Chorus

Session setup, Pierre Boulez conducting the BBC Symphony Orchestra and Chorus at the Methodist Central Mission in West Ham, London, in 1974

Session setup, Pierre Boulez conducting the BBC Symphony OBoulez rchestra and Chorus at the Methodist Central Mission in West Ham, London, in 1974

Because this was a recording for surround quadraphonic sound, it was necessary to rearrange the orchestra in such a way that certain sections could be ‘folded’ round to the back speakers in the final four track remix, and I am grateful also that Pierre Boulez agreed to face an unconventional orchestral setup. Although the listener hears the work in surround sound, I believe it is wrong to expect the conductor to manage a ‘surround’ orchestra (although it has been done, with many accompanying headaches). Therefore, borrowing a leaf from pages of pop recording sessions, we worked on the principle that loud instruments should not play into the microphones being used by soft instruments. (I remember seeing many pop orchestra sessions, in which the violins sat at the back, winds in front of them, and brass in the front, to achieve the same effect of separation.)

We were not looking for isolated separation in our recording, since a good classical sound depends upon a certain amount of cross-channel leakage, but enough presence and ‘accent’ on each individual track to permit us to relocate the instrumental sections later. Therefore, we arranged the orchestra in a semicircle, with the woodwinds on the outside left, the brass on the outside right, the strings and harps in the center, the horns (blowing backwards!) to the rear left, with the timpani and percussion spread across the back of the hall. The choir was placed in the balcony above the orchestra. Our intention was, therefore, to have woodwinds, horns and timpani coming from the rear left speakers, and brass and percussion from the rear right; the orchestra was laid out in such a way that the natural leakage of sound between one section and another would only help to create a fully circular rather than ‘cornered’ effect. We did not listen to playbacks at the sessions in quadraphonic sound, but used four kWh speakers, strung across the front wall in a straight line. I must also mention the soloists, who were placed behind the conductor, facing the orchestra. This allowed them a certain degree of separation and certainly avoided too much orchestral leakage into their microphones during the larger musical climaxes.

Having written so much about the preparations for the recording, I must add the somewhat anticlimactic note that the sessions themselves went extremely smoothly and without any great problems. It took the conductor some time to adjust to the new balance of orchestral sound, especially the very large brass section playing in his right ear and without the usual benefit of distance to reduce their volume. The singers: Jess Thomas, Marita Napier, the superb Yvonne Minton, Kenneth Bowen, Siegmund Nimsgern and Günther Reich (who is both speaker and, for a few, wonderful notes, singer) performed supremely well. Each song/section was recorded complete, to maintain the musical flow of the work, and any retakes were usually complete. It might be interesting to note that Jess Thomas and Marita Napier (the ill-fated Waldemar and Tove who are the Romantic lovers of the piece, despite the fact that they never sing together) could not attend sessions during the same periods and, thus, never actually met throughout the recording of the work.

We encountered problems in only one are: with the choirs. In both the male chorus of Waldemar’s Vassals and the final mixed chorus, the entire orchestra is  called upon to play at full force, and the resulting sound was so gigantic that we had the greatest difficulty in picking up the chorus, despite its size. This was caused partly, I believe, by the reverberation of the building itself and, no matter how one tried to mask the choral microphones, gigantic attacks by the brass and percussion swamped them. During the first choral recording session, the chorus mikes were slung across the hall without even the mss of the balcony to mask them from the orchestra, and the results were disastrous, with the voices barely audible. Later, in the second choral session, the microphones were moved behind the rim of the balcony, with better results, even though some of the orchestra tended to seep through the structure of the building, interfering with the vocal effects. This proved to be a slight problem in the stereo remixing stage, but was surprisingly much easier to accommodate in the quadraphonic remixing. Those listeners who have a chance to compare the stereo record with the quadraphonic will perhaps be surprised by the additional power and volume of the chorus in quad, possibly because it is spread, between back and front, on either wall. Nevertheless, the problem was minor. I am happy with the overall result but, like any producer, could always have used that extra ounce more.

So, Schoenberg’s Gurrelieder has been recorded, mixed, but (excellently!) and pressed. In the monthly release lists, it is just another entry and, since those sessions, Bob Auger and I have worked together on Schoenberg’s opera Moses and Aaron and on two albums of operatic arias by Renata Scotto (with two different orchestras in two different churches). Bob has had many recording sessions with other producers, and I have worked with several other engineers on any number of projects, ranging from solo guitar to a full-length opera. That is the way our work goes, but I think we will always remember the Gurrelieder sessions with special challenge in the recording of a unique work. Hopefully, if you have read this far, you will listen to the finished records, and judge the results for yourself.

– Paul Myers

Part Two: Engineering
For many years as a young and enthusiastic music lover, I discovered all kinds of learned books and journals (all beyond my comprehension) on orchestration and music appreciation. Practically all these worthy tomes had one thing in common however – when the authors wanted to mention a vast orchestral work as an example it was always Gurrelieder. In these days (during and just after the war) you had no chance of hearing what the writers were driving at so you had just to take the comments for granted and gradually become overawed at the potential of a piece of music which has never heard.

I didn’t’t have the opportunity to hear the Prom performances Paul Myers referred to earlier and I decided not to listen to the Deutsche Grammophon discs when I was first approached by CBS to engineer the sessions in case some kind of prejudice set in; consequently I faced the first session with some apprehension. Some later Schoenberg I knew, but apart from the beautiful Verklärte Nacht I had no experience of his youthful work. I had a feeling that it was some sort of super-Wagner or even Mahler but that was all. These feelings were not dispelled when Paul sent me the orchestral and choral lineup to dwell on.

Having recently discovered the West Ham Central Mission from a recording point of view, I asked Paul to come and visit some sessions I was recording for another customer (always a delicate matter to suggest to both partners). I don’t think I’ve ever seen a major decision made quite so rapidly. Immediately, on entering the auditorium, Paul said: ‘This is the place; it feels absolutely right’. We were, at that time, recording some reasonably avante garde orchestral pieces but, even though this was very different from the task to follow, Paul had no qualms about his decision.

There were many details to be taken care of and (as I never tire of reiterating in regard to recording on location) 50% of the problems occur before a single foot of tape passes the record head. The sheer physical problem of accommodating all those people for long stretches of time have to be considered: food, toilets, transport, heating and lighting all have their priority and are all problems concerning musicians and singers which are liable to be thrown at a busy engineer without warning. There is nothing more distracting than an irate clarinet player who hasn’t’t got enough light when you are wondering why you can’t hear the concertmaster properly in his solos when the nearest microphone is only five feet away.

The actual placement of the orchestra and chorus in the church was carefully worked out in advance, and I must say that at this stage a lot of faith was placed in the directional capabilities of some of the microphones (particularly in regard to the surround pickup for the quad mix).

One or two of the microphones were switched to a ‘figure-of-eight’ characteristic since there is ‘nothing so dead as the side of a figure-of-eight mic’. The stray pickup from the rear of the mic is less troublesome than the spurious pickup of next-door instruments; in particular, I refer to the possibility of picking up the clarinets on the first violin microphones and the bass on the cello microphones. As Paul has said, the separation available on the quad mix was most surprising and allowed us to place the instruments comfortably round to their session layout.

The drawing gives some idea of the layout and it will be seen that the resultant quad mix merely takes the listener into the orchestra to a position held by the conductor on the session, except for the fact that it was decided to keep the soloists in front of the listener and obviously altered the heard balance to some extent.
The problem of getting enough chorus sound over the huge orchestra was real enough. I doubt if they can be heard properly in concert but of course that is no criteria for recording, when the sound quality has to make up for the loss of visual impact. There were a few worrying moments here but, I think, in the end they were satisfactorily overcome – at least most of the reviewers have been kind enough on this detail.

Once the overall sound and balance had been arrived at on the first session, considerable thought was given to the problem of dynamic range. It is now possible for the professional engineer to set his recording to accommodate the loudest sections of the music and let the quieter sections look after themselves.
This is not so bad with high quality monitor systems but the prime aim is to produce a tape suitable for transfer to disc and domestic equipment. Consequently, some gentle manual squeezing of the dynamics took place on the session and on the resultant mix down. You can only have 100% of volume at any one time and how this 100% is shared out among 500 performers is a major problem, particularly when you consider that a solo pianoforte recording uses 100% for one musician most of the time. It’s all a matter or perspective and I trust Gurrelieder doesn’t sound too much as though looking through a landscape through the wrong end of a telescope.

It is not generally known that the Gurrelieder sessions were interrupted by the recording of the complete opera Moses and Aaron by the same composer. The only real problem for the engineers here was the fact that Paul decided to record this work with the orchestra turned through 180° compared with Gurrelieder. Sufficient to say that the last sessions of both works were carried out on the same day – but that’s another story.

– Bob Auger