Chapter 6
Orchestral rooms
Orchestral music was designed to be performed live. When many of the great classical works were written there was no such thing as recording, so the instrumentation and structure of the music was aimed only at its performance in spaces with audiences. Transferring the performance into a studio, perhaps of only just sufficient size to fit the whole orchestra, imposes a completely different set of conditions. As was discussed in Chapter 4, a big constraint on the achievement of a natural orchestral sound in studios is the fact that the whole process is usually entombed in a massive acoustic containment shell. This removes even further the ambience of the studios from that of the concert halls.
Today, much orchestral recording is for the soundtracks of films, and in such circumstances, when the conductors need to see the films as well as to be in close contact with the musicians, then the facilities of a studio are more or less mandatory. On the other hand, under less technically demanding circumstances, ever since the earliest days of recording, a large proportion of orchestral recordings have been made outside of purpose-built studios. However, when one begins to consider the deeper element of achieving good orchestral recordings, the above fact comes as no surprise.
6.1 Choice of venues and musicians’ needs
Around the world there are some very famous and widely used locations for orchestral recording. Not too surprisingly, concert halls are one member of this group, but assembly halls, churches and cathedrals are also popular locations. One requirement of such a space is that it needs to be big enough to house the orchestra, but usually, the apparent acoustic space needs to be even larger.
If we first consider the obvious, recording in a concert hall, there are two conditions likely to be encountered; recording with or recording without an audience. When an orchestral recording is being performed for recording purposes, in almost all cases, the conductor will want to discuss the music and his interpretation of its performance with the musicians. It may be necessary to run through separate sections until the right feel is achieved, and errors or misinterpretations in the playing may be pointed out and corrected. Performances intended for recording tend to be auditioned in greater detail than live performances, and a small detail which may pass in a live performance may become irritating on repeated listening to a recording. The straightening out of these small points can be very time consuming. Clearly, much of this is private and personal, and it is usually all conducted in an atmosphere of professionalism, trust and understanding. Frank and open discussion of these points would rarely be possible in front of an audience, nor would it be likely to constitute an interesting spectacle for the paying public, so almost out of necessity, many such recordings take place out of public view. Unfortunately, if such recordings are made in a concert hall, then that hall will probably have been designed to have an appropriate reverberation time when full. Its empty acoustic may not have been a prime object of its design, although, in an attempt to make the acoustics independent of audience numbers, some halls have empty seats with absorption coefficients which try to match that of an occupied seat.
With recording techniques using close microphones this may or may not be of great importance, but for recordings with more distant microphones, the ambience of some empty halls may be undesirable. However, surprising though it may seem, even when full, not all concert halls are highly rated for orchestral recording purposes. A concert hall is an expensive thing to build, and few can be dedicated solely to orchestral performances. Consequently, ideal orchestral acoustics may have to be compromised to accommodate other uses of the halls, such as for conferences, operas, electrified music concerts or jazz performances. Each use has its optimum set of acoustic conditions, both in terms of hall acoustics and stage acoustics. Furthermore, in addition to the optimum reverberation requirements, lateral reflexion characteristics may also form another subdivision of the ideal acoustics for each use.
Some of the old halls, built before recording existed, are still very well liked. In fact, this is not too surprising, for they frequently did not have to make quite so many performance compromises as halls of more recent construction. Moreover, when much of the classical music repertoire was written, it was written with many such concert halls in mind. The fact that much classical music can be expected to sound good in those halls is something of a self-fulfilling prophesy. Live performances were the only performances to be heard by the public when those halls were built, and no compromises needed to be made to allow for electric amplification or for other uses. Many orchestral pieces were even written with specific concert halls already considered for their first public performances. However, this is somewhat akin to writing and recording a piece of music in a studio with idiosyncratic monitoring: it may not easily transfer to more acoustically neutral surroundings.
The great composers usually also had a comprehensive understanding of the needs of the musicians. Orchestral musicians need to hear themselves in a way that is both clear and sonorously inspiring. They also need to hear clearly many of the other musicians in order to develop the feel of the performance. Composers often bore these facts in mind when arranging the instrumentation, so the music, the instruments themselves and the halls in which they frequently performed developed not in isolation but in concert with each other. It is thus little wonder that many of the shoe-box shaped halls, which have been commonplace for centuries, are still well liked by musicians, recording personnel, and audiences alike.
In recent years it has become apparent that, for a rich sense of spaciousness, lateral reflexions are of great importance. The shoe-box shaped designs provide plenty of these, but they can be very problematical for many other purposes for which the hall may be used, often playing havoc, for example, with the intelligibility of the spoken word. Conferences or speeches in such halls can be nightmares. In fact, the reason why so many religious masses are chanted and not spoken is because the longer sounds of a chant are less easily confused by the reflexions and reverberations of most churches than would be the more impulsive consonants of normal speech. Hard constants would tend to excite more modal resonances, as they contain more frequencies than the softer chanted consonants. Ecclesiastical chanting therefore appears to have been of acoustic origins rather than religious ones.
Nowadays, however, we must perform much of the old music in multifunctional halls, or in churches, town halls, and the like. What all of these locations have in common though, which sets them apart from most recording studios, is the large amount of space which they have for the audience or congregation. They also tend to have many windows and doors, which allow an acoustic coupling of the internal spaces to the outside world. They are not constrained within the bunker structure of a sound isolation shell, and all, therefore, have room to ‘breathe’ at low frequencies; even if this is not immediately visually apparent from the interior dimensions of the rooms themselves. Given this acoustic coupling through the structure, they tend to appear to be acoustically larger than they are physically.
The optimum reverberation time (RT) for orchestral recordings tends to vary, according to the music and instrumentation, from about 1.8 to 3 seconds in the mid-band, with the 5 kHz RT usually being around 1.5 seconds. The low frequency RT is a continuing controversy, with differences of opinion as to whether the 100 Hz RT should rise or not, and if so, by how much. In general, there are compromises between definition, warmth and majesty. As the RT rises definition is lost to warmth, which in turn changes it to a character of majesty, until at higher RTs, all is lost to chaos. Inevitably the optimum requirements for each piece of music or each orchestral arrangement will call for somewhat different conditions, as will the presence or absence of an organ, chorus, solo piano, or number of contrabasses in the orchestra. It has certainly been my experience that a moderate rise in RT at low frequencies is usually desirable; but that introduces yet another variable – personal preference.
It was pointed out to me some time ago by the New York acoustician Francis Daniel that the Fletcher-Munson curves of equal loudness could have something to do with the LF debate. Figures 6.1 and 6.2 show the classic Fletcher-Munson curves, alongside the more modern equivalent, the Robinson-Dadson curves. The latter are now generally accepted as being more accurate; however, the differences are too small to have eclipsed the former in general usage. What they both show is how the sensitivity of the ear falls at the extremes of the frequency range. If the 0 dB threshold line at 3 kHz is followed to 30 Hz, it strikes the sound pressure line at 60 dB. If the line passing through the 25 dB point at 3 kHz is followed down to 30 Hz, it will be seen to pass through the SPL of around 65 dB. These are curves of equal loudness, and the above observations mean two things. Firstly, that 60 dB more (or one million times the acoustic power) is needed at 30 Hz to reach the threshold of audibility than is needed at 3 kHz. An extra 40 dB (or 10,000 times the power) is needed at 30 Hz to sound as loud as a tone of 25 dB SPL at 3 kHz. So it can be seen that the ear is vastly more sensitive to mid-frequencies than to low frequencies at low sound pressure levels. Secondly, the rise of 25 dB in loudness from 0 dB SPL to 25 dB SPL at 3 kHz needs only a 5 dB increase at 30 Hz to produce the same subjective loudness increase.
Figure 6.1 The classic Fletcher and Munson contours of equal loudness for pure tones, clearly showing higher levels being required at high and low frequencies for equal loudness as the SPL falls. In other words, at 110 dB SPL, 100 Hz, 1 kHz and 10 kHz would all be perceived as roughly equal in loudness. At 60 dB, however, 10 kHz and 100 Hz would require a 10 dB boost in order to be perceived as equally loud to the 1 kHz tone
Looking at these figures again, 25 dB above the threshold of hearing at 3 kHz will sound as loud as 5 dB above the threshold of hearing at 30 Hz. The perceived dynamics are considerably expanded at the low frequencies. At high SPLs above 100 dB, the responses are much more linear. Now let us consider a symphony orchestra in a hall, playing at 100 dB. The direct sound will, according to the 100 dB lines on the equal loudness curves, be perceived in a reasonably even frequency balance. However, when the reverberant tail has reduced by 50 dB, and is still very clearly audible at 50 dB SPL in the mid-range, the lower octaves will have fallen below the threshold of audibility.
Considering the fact that much of the reverberation that we perceive in concert halls is that of the tails, after the effect of the direct and reflected sounds has ceased, then much of our perception of that reverberation will be in the area where the low frequencies would be passing below audibility if a linear reverberation time/frequency response existed. This suggests that halls with a rising low frequency reverberation time may be preferable for music which is performed at lower levels, in order to maintain a more evenly perceived frequency balance in the reverberation tails. Rock music on the other hand, played at 120 dB in a hall, may well degenerate into an incomprehensible confusion of low frequency wash in halls in which the RT rises at low frequencies. In fact, electrified rock music usually requires a much lower overall RT than orchestral music, and, quite definitely for rock music, a more linear RT/frequency ‘curve’ is desirable, or even one where the low frequency RT falls.
Figure 6.2 Robinson-Dadson equal loudness contours. These plots were intended to supersede the Fletcher-Munson contours, but, as can be seen, the differences are too small to change the general concept. Indeed other sets of contours have subsequently been published, as further updates, but for general acoustical purposes, as opposed to critical uses in digital data compression and noise-shaping, the contours of Fig. 6.1 and the above suffice. The lower MAF (minimum audible field) curve replaces the ‘0 phons’ curve of the older, Fletcher-Munson curves. The MAF curve is not absolute, but is statistically derived from many tests. The absolute threshold of hearing varies not only from person to person, but with other factors such as whether listening monaurally or binaurally, whether in free-field conditions or in a reverberant space, and the relative direction of the source from the listener. It is therefore difficult to fix an absolutely defined 0 dB curve
As has probably already been gathered from all of this, to produce a large single recording studio to mimic all of these possibilities would be a mighty task indeed. It is one very significant reason why so many orchestral recordings are done in the aforementioned wide range of out-of-studio locations.
Live performances inevitably must be recorded in public halls, though the reasons for recording live are not always solely for the hall acoustics. It can be very difficult to lift the definitive performances out of orchestras in studios. The adrenaline of a live performance will, in most cases, give the recording an ‘edge’ which the studio recordings frequently fail to achieve. Nevertheless it is still heard from musicians that far too many studios also fail to inspire a performance from an ambient point of view. All too often, the ‘technical’ environment of many studios is not conducive to the ‘artistic’ feel of a hall. Comfort and familiarity with the surroundings can be a very important influence on musicians, as can their own perception of their sound(s). In most concert halls, or assembly halls and such like used for recording, the performing area is usually surrounded on at least three sides by reflective surfaces. As mentioned before, these lateral reflexions are of great importance to the sense of space, and by association, with the sense of occasion. If the acoustic conditions in a studio can help to evoke the sense of the occasion of a concert stage, then it may well be off to a good start in terms of the comfort factor for musicians.
The one big obstacle in re-creating realistic performance-space acoustics in a studio is the normal lack of the open space in front of the orchestra. During concerts, the musicians face from the platform into a hall from which there will probably be no significant reflexions, but much reverberation. The musicians thus receive the bulk of their reflexions from the stage (platform) area, and the subsequent reverberation from the hall in front of them. Human hearing is very directional in the horizontal plane, so the perception of the necessary reflexions, which help to reinforce and localise the instruments, can be easily discriminated from the equally necessary reverberation. The reverberation does not swamp the reflexions because of the spacial separation of the two.
In a studio of practical size, we have a great problem in trying to re-create this situation. If the orchestra face a reflective wall, the reverberation will be confined within the area of the performance, and additional reflexions will also arrive from the front wall. If we make the wall absorbent, we will tend to kill the overall reverberation. Either result is unnatural for the orchestra. It seems to me that perhaps the only way to overcome this would be with a relatively absorbent wall containing an array of loudspeakers, coupled to a programmable artificial reverberation system. This way a reasonably sized studio could be created, say about the size of a concert platform, but with the effect of a reverberant hall in front of the musicians. The directional characteristics of the sound could thus be preserved whilst keeping the studio to a reasonable size.
It is doubtless the directional sensitivity of our hearing which is at the root of the phenomenon that when using conventional microphones (SoundField microphones can be an exception to this) the necessary microphone position for a subjectively most similar direct/reflective/reverberant balance is usually considerably closer to the orchestra than would be the corresponding seating position for a listener. However, it must be said that the microphone positions are usually above the audience, so they perhaps suffer less from local absorption, but this does not account for all of the effect.
What tends to set orchestral studios apart from most other studios is that a generally rectangular acoustic shape is rather more common. Walls are normally faced with a significant amount of diffusive objects to break up unwanted cross-modes, but the more or less parallel-sided nature of the walls can help greatly in the reproduction of the environment of the platform of a hall. Ceilings, though, as in halls, are almost never parallel to the floors. In some multi-functional rooms, they may be found to be parallel, but they are most likely to be relatively absorbent. Many concert halls have acoustic reflectors above the platform, usually angled to project more of the sound towards the audience, though frequently the sound also rises into scenery and curtain raising mechanisms, which can be absorbent in their overall effect.
However, in no cases can studio acoustics be separated from the polemic on the subject of the use of close versus distant microphones. I personally have many dozens of classical CDs, and must admit in many cases to enjoying the dynamism of the close mic’d recordings when I am listening to them, yet I can equally enjoy the more integrated grandeur when I listen to many of the ‘stereo pair’, distant mic’d, or SoundField recordings. The problem is, of course, that a listener at a live concert cannot be both near and far from the orchestra at the same time. With microphones, via a recording, it is possible, but to my ears the result is perceptually confusing unless either one is greatly subservient to the other, and present only for added detail or richness. The problems of which option to take usually revolves around so many of the fundamental problems of psychoacoustic perceptions, some of which are mutually exclusive. Essentially, close and distant microphone techniques are two entirely different things, and the use of either is a matter of choice. They are not in competition for supremacy, but merely options for the producers, who will choose the technique which is considered to be appropriate for any given occasion.
6.4 Psychoacoustic considerations and spacial awareness
Not all of the problems of the design of recording studios for orchestral use are of direct concern to the recording engineers. Many other things which affect only the comfort and sense of ease of the musicians must be taken into account if the best overall performances are to be recorded. Sometimes these things can be better explained by looking at extreme situations, which help to separate any individual characteristics which may only be subconsciously or subliminally perceived in the general confusion of the recording spaces. So let us now take a look at some of the spacial effects which can play such great parts in the fields of awareness and comfort.
I remember my first visit to the Anglican Cathedral in Liverpool, UK. I went to hear its pipe organ, which was originally commissioned in 1912 as the largest and most complete organ in the world. It has 145 speaking stops, over 30 couplers, around 9700 pipes, and on Full Organ can produce a 120 dB SPL in the 9 second reverberation time of the cathedral. It is blown by almost 50 horsepower of blower, and it needs it all, because the cathedral is one of the ten largest enclosed spaces in the world. Its size is truly awesome, and a well-known producer accompanying me at the time admitted to being frightened by its overwhelming size.
When inside the cathedral, one is aware of a reverberant confusion from a multitude of noise sources, but occasionally, on windless days, and when not open to the public, there are occasions of eerie silence. When speaking in a low-to-normal voice to a person close by in the centre of the great nave, it is almost as though one is speaking outside in a quiet car park. It is not anechoic, as there is a floor reflexion to liven things up, but the distances to the other reflective surfaces are so great that by the time they have returned to the speaker, they are below the threshold of audibility. On the other hand, a sharp tap on the floor with the heel of one’s shoe produces a sharp rap, followed almost half a second later by an explosion of reverberation. The clarity of the initial sound is absolute, as coming from the floor, it has no floor reflexion component; yet a subsequent tap, originating before the reverberation of the first one has died down, is almost lost in the ambient sound. This highlights well the temporal separation discrimination which plays so much part in orchestral acoustics.
Back in 1982, at a time when I was still involved in the recordings of Mike Oldfield, I was visited by Hugo Zuccarelli, the Argentinian-Italian ‘Holophonics’ inventor. He came to a studio where I was recording to see if I thought that Mike would be interested in using ‘Holophonics’ on his recordings. Mike never used it, but Pink Floyd and others certainly did. On headphones at least, Hugo could pan things around the head with stunning realism; and a realism not only of position but of clarity as well. He also had a demonstration cassette of rather poor quality, yet the sounds of jangling keys remained crystal clear in front or behind the head, whilst the tape hiss remained fixed between the ears. It was an impression that I will never forget, as the wanted sounds had become absolutely separated from the noise, existing totally natural and noise-free in their own spaces. This demonstrated most clearly to me the powerful ability of the ear to discriminate spacially, even in what amounts to an entirely artificial environment. The holophonics signal supplied strong phase-/time-related signals to the ears, not present in mono signals, nor necessarily in multiple microphone stereo recordings. The extra phase/time information provides a powerful means for the brain to discriminate between different signals, including between wanted and unwanted sounds.
The old chestnut of the ‘cocktail party effect’, to which the holophonics effects closely relate, has been around for years, and is no doubt already familiar to most readers. To recap, if a stereo pair of microphones is placed above a cocktail party, and auditioned on a pair of headphones, but panned into mono, then the general murmur of the party would be heard, but it would be difficult to concentrate on any individual conversation unless one of them happened to be occurring very close to the microphones. However, when auditioned in stereo, at exactly the same level, suddenly it becomes clear that many separate conversations can be easily recognised and understood. The precise audiological/psychoacoustical mechanisms behind this have still not been fully explained, but nonetheless the effect itself is very well established as fact.
If we lower the level of the cocktail party recording into the domain of the background noise of the tape hiss, then the conversations will begin to get lost in it, masked by the hiss even whilst still relatively well above the absolute hiss level. But such is not the case with the effect of holophonics, which can lift a recognisable pattern out of the hiss by spacial differentiation. The phenomenon, once recognised, has a lasting effect. The significant difference between the effects of holophonics and conventional stereo is that conventional stereo only provides information in a single plane; the holophonics effect is three-dimensional. The single plane of conventional stereo contains all the information, both wanted and unwanted, so the tape hiss shares the same sound stage as the wanted signals: they are all superimposed, spacially. In holophonics, however, the wanted signals can be positioned three-dimensionally, yet the tape hiss is still restricted to its single plane distribution. Once sounds are positioned away from the hiss plane, they exist with a clarity which is quite startling, and it is remarkable how the hiss can be ignored, even at relatively high levels, once it is spacially separated from the desired signals.
We thus have several distinct mechanisms working in our perception of the spaces that we are in, and all of the available mechanisms are available to the musicians in their perceptions of the spaces in which they are performing. Clearly, all of these mechanisms cannot be detected by microphones, nor can they be conveyed to the recording medium, and consequently, many of them will never be heard by the recording engineers in the control rooms. In fact, although the recording engineers may well, consciously or subconsciously, perceive these things, they often fail to fully realize their importance to the musicians. To the musicians, however, who work with and live off these things daily, their importance cannot be over-stressed. In the recording room therefore, all of these aspects of the traditional performance spaces need to be considered. At the risk of becoming repetitive, I say again, if the musicians are not at ease with their environment, then an inspired performance cannot be expected.
From time to time, there are situations which require even orchestral musicians to wear headphones when recording; either full stereo headphones, or single headphones on one ear. Perhaps the major reason why this is usually avoided is not the enormous number of headphones required, but that inside headphones, certainly if they are of the closed type, it is not possible to reproduce the type of complex sound-field that the orchestral musicians are accustomed to playing within. Human aural pattern recognition ability is very strong, often even when sounds are hidden deep within other sounds, and even deep in noise. If headphones isolate the musicians from the usual patterns, the effects can be very off-putting.
For many of the reasons discussed in this chapter, orchestral studios should not be designed primarily simply for what is good from the point of view of the recording personnel. First and foremost, a performing environment should be created to aid the whole recording process. I realise that there are hi-fi fanatics who disagree, but for me, I would much rather hear a compromised recording of an excellent performance, than an excellent recording of a compromised performance. Obviously, though, the real aim of the exercise is to make excellent recordings of excellent performances, and this is why so many things must be considered.
I must admit that I did not fully appreciate the importance of the recording environment, or the extreme fussiness of some musicians about their foldback balance, until, when producing a recording, I was having problems getting some backing vocalists to sing with the appropriate phrasing. I was still unhappy with the performance after they had left, and was singing along with the tape, showing the engineer what I meant, and seeing if he agreed. He not only agreed, but said that my voice also suited the part, then encouraged me to go into the studio to try to see how it fitted the track. In the space of minutes, foldback had suddenly, for me, become the most important thing in the whole process. If the overall sound was too loud, I could not phrase properly, if it was too quiet, it was difficult not to swamp it by my own voice inside my head. If my voice was too low in the track, I strained, and tended to sing sharp; and if it was too high in the track, I held back, lost my dynamics, and had a tendency to sing flat. From that day onwards, I suddenly saw foldback in a very different way. The experience was quite a shock, and I felt a terrible guilt about the many times in the past when I had perhaps been a little less considerate to the musicians on that subject than I should have been.
Many musicians of all kinds play off their own sound; it is like a feedback mechanism that both reassures and inspires them. This is true of more or less every musician, and of whatever type of music that they perform. If their sounds are not given to them in their foldback as they need, then it can be very disconcerting. In the case of orchestral musicians however, their foldback is not usually via headphones, but via the reflective surfaces of the room. If playing to a backing track, it will possibly only be the conductor who will listen to it on headphones. In these cases, the acoustics of the performing space provides the foldback to the musicians, so it should be given its due consideration as such.
Another frequent conflict between the needs of musicians and those of the recording engineers is the use of acoustic screens. In the close-microphone type of recording, separation is a factor which is often considered desirable by the engineers. This can be improved by the use of acoustic screens between different sections of instruments, which normally, as a concession to the musicians, have windows to allow eye to eye contact. Unfortunately, this can disrupt the perception of the desired acoustic sound-field by the musicians, and in most cases, they would prefer the screens not to be there.
It has been quite amazing how, over the course of recording history, musicians’ needs have often been neglected in the recording process. Time and time again, if the recording engineer has had a problem with overspill, screens have been imposed upon the musicians, without due consideration of the effects of their insertion on other aspects of the recording process. In many cases, recording staff have totally failed to appreciate the artistic damage which can be done by delays and disturbances caused by technical adjustments. Awareness of such things has been one thing which has set apart the specialist recording engineers and producers, who by understanding things in a more holistic way, have gained the co-operation and respect of the musicians, and have thus produced more inspired recordings from which they have deservedly built their reputations.
This discussion on some rather peripheral aspects of audiology and psychoacoustics has not been a digression, but is a fundamental requirement in understanding what is necessary for the design of good recording spaces for orchestral performances (or perhaps rather, for the design of good performance spaces for orchestral recordings). What must have become evident from the discussion is that acoustic variability is a prime consideration in any such design, unless the space is to be restricted to excelling in the recording of perhaps only a small range of the likely performances for which it may be required.
The room variability needs to be in terms of overall reverberation time, and if possible, the relative balance of low and mid/high frequency reverberation. Reflexions need to be controllable in terms of time, direction, and density, with the availability of some reasonably diffusive surfaces to add richness without undue colour. Always, however, the consideration of what the musicians need should be given at least equal weight to the needs of the recording staff when acoustic adjustments are being made. It is very necessary to strike a balance between these priorities, and to achieve the close cooperation between all parties involved. Always remember, the variable acoustics are not just provided for the benefit of the recording staff.