Loudspeaker Orchestras

Many musicians spend their lives trying to cram hundreds of tracks into just two speakers — but performing with an NSML system presents exactly the opposite challenge!

The BEAST system being set up before a concert.Photo: Jonty Harrison

We tend to think of loudspeakers as devices that just make music louder. But what if the loudspeakers themselves were part of a musical performance? That's the idea behind non‑standard multi‑loudspeaker diffusion systems, where diverse loudspeakers are placed everywhere across the concert room and wired to the bus outputs of a control surface or a mixing desk. In these setups, a performer directs recorded sound around the speakers, treating the array as if it were a concert‑hall‑sized musical instrument. The result is usually quite fascinating: the music surrounds the audience, moves around and keeps changing colour and shape.

These unorthodox methods have been gathering hardcore fans around the world for a good 50 years. In this article, we'll meet some of the people using them, look at what's involved from a technical point of view, and find out what sort of 'performances' they make possible.

Starting Out

The first non‑standard multi‑loudspeaker diffusion systems (or NSML systems, for short) were developed in San Francisco and in Paris, France, in the '60s. On the West Coast, composer Stan Shaff and equipment designer Doug McEachern introduced their Audium system at the University of California in 1960, settling into their own theatre in 1968. Meanwhile, at the Groupe de Recherche Musicale in Paris, engineer, inventor and composer Pierre Schaeffer developed research work he had begun before World War II. One result was the Acousmonium, a loudspeaker orchestra that was made fully functional by composer François Bayle in 1974.

NSML systems made it possible to include space as a full dimension of musical composition. As Stan Shaff puts it, when you listen to a performance at the Audium Theatre, "sounds are 'sculpted' through their movement, direction, speed and intensity on multiple planes in space”. They also provided an answer to the age‑old problem of how to create engaging live performances of electronic music.

At the time, electronic music was still a minority interest and, as a consequence, NSML systems became associated with 'avant garde' music. Unfortunately, the growth in popularity of electronic music in the '80s didn't change this, and the vast majority of performances based on recorded music in techno, trance, and so on are played on stereo setups, while NSML systems are still used mainly to perform 'avant garde' music. This is a pity, because NSML systems have much to bring to any kind of music.

NSML systems in Europe include the BEAST (Birmingham ElectroAcoustic Sound Theatre) system at the University of Birmingham, which is run by Jonty Harrison, while in Paris, Vincent Laubeuf is the head of MOTUS, a privately held company that specialises in concerts involving non‑standard multi‑loudspeaker setups. Jonty Harrison manages the BEAST multi‑speaker system at the University of Birmingham.Vincent Laubeuf of MOTUS.

Performance

Central to the idea of NSML systems is the notion of performance, albeit of a particular kind. During these concerts, a large number of loudspeakers are generally set up. Jonty Harrison says he usually uses between 70 and 90 speakers during BEAST concerts, and the MOTUS equipment list gives the client the opportunity to use a 100‑loudspeaker system.

The art of performing on such a system lies in taking advantage of multiple factors such as speaker placement, speaker colour, and room acoustic properties. To fully take advantage of those factors, NSML systems include routing matrices that enable any channel from the playback medium to be directed towards any loudspeaker, as shown on the diagram. According to Jonty Harrison, "You want to enhance the musical shapes and gestures that are already contained in the music, by articulating the space with the musical material. You may, for instance, translate dynamic shift or energy to spatial shift or energy.” Synopsis of the MOTUS routing system. Notice the frequency‑selective amplification.Photo: Jonathan Prager

Translated into practical terms, this means that the beginning of a musical crescendo might be played on a limited number of loudspeakers, and the louder the music gets, the more loudspeakers would be used. Similarly, a thin, mono musical part in a mix would be played on a given, narrow zone from the loudspeaker array, whereas a large original stereo image would be dispatched on a wider choice of loudspeakers. Staying faithful to the music, while enhancing the way it's written: this is indeed performance.

In music, generally speaking, performance requires knowledge of both the piece to be played and the instrument. This is also the case with NSML systems: live shows have to be carefully rehearsed so that the performer gets to know the music to be played, the loudspeaker setup, and its interactions with the concert-room acoustics. Such knowledge is of paramount importance. For instance, according to Vincent Laubeuf, it's part of the MOTUS performance vocabulary to be able to play medium‑high frequencies on dedicated speakers close to the audience. But this must be done properly: how does that sound from the audience's point of view? Is there a part of the music piece that's particularly well suited to this particular colour? Are those loudspeakers going to be used as support, or as main sources? What about the other frequencies that are heard at the same time? How does this work with the adjacent parts of the music? Many decisions have to be made, and this can prove disorienting at first, especially when one is dealing with large and complex setups such as the one shown in the diagram.

Sound In Space

Speaker setup for a BEAST concert in Elisabethkirche, Berlin: a complex instrument!Photo: Jonty Harrison

An important question of NSML system diffusion concerns how many tracks of recorded music to start with. While it might seem a bit strange to play stereo material on a 100‑speaker system, it's actually perfectly feasible. Vincent Laubeuf even prefers it over playback of multitrack material: in his opinion, having a very limited number of original sources being played simultaneously via a large number of loudspeakers often gives better results than having many speakers behaving independently. This is all the more true on the MOTUS system, which is based on extremely dissimilar loudspeakers. As for Jonty Harrison, though he understands Vincent's point of view, he personally prefers to work with a higher number of tracks (his own recent work is typically eight‑channel), for more flexibility.

Whether one chooses to perform from traditional stereo content or from multitrack material, the issue of spatial perspective necessarily arises. For instance, in mixes destined for stereo playback, a sense of distance is typically obtained by adding reverb and/or filtering the high and low frequencies. This creates a virtual image of a 'faraway' musical object, which will seem to be emitted from way behind the speaker position. In NSML diffusion, 'faraway' will be more literal, since it can be obtained simply by playing that particular sound on distant loudspeakers. Also, in stereo, groups may be mixed on different 'zones' of the stereo image. In NSML diffusion, this will be translated to moments or tracks being directed towards different groups of speakers.

This photo shows the elaborate rigging necessary to position the speakers within the Elisabethkirche.Since there are no instrumental performers to look at, and no stage, should the traditional 'front‑back' / 'left‑back' concert space organisation be abandoned in favour of a more flexible, 'isotropic' setup? At the beginning of MOTUS, traditional concert directivity was altogether forsaken, but Vincent Laubeuf says that it was gradually reintroduced: listeners appear to pay more attention and feel better when most of the sonic information comes from the front. Jonty Harrison's opinion is that isotropy may help one person who's completely out of range, but generates a lower general audio quality. Also, he feels it may distort the original composer's intentions, by restraining the possible range of musical expressions: for instance, there is no possibility of 'intimacy' if the distance between the listeners and each speaker is random. He also states that this debate is often a point of disagreement between him and his students. In my opinion, a major point in favour of isotropy would be that it makes the notion of sweet spot disappear altogether. Listeners are 'inside' a sound scene, and can watch it from any point of view they want, even moving from one point to another during the performance. The matter seems open to debate...

Colouring

A central issue when performing with an NSML system is the coloration of the original musical content. Understandably, playback on such systems is not like playback on professional‑range headphones: first, the sound will be coloured by the loudspeakers, and then by the room acoustics. NSML system owners can adopt different attitudes towards this phenomenon. According to Jonty Harrison, trying to remain true to the original colour of the recording is most important. This leads BEAST to use as many full‑range loudspeakers as possible, so as to minimise the coloration. As for reverberation, it is considered a component of space, thus being integrated into the spatial image. At MOTUS, Vincent Laubeuf handles things a bit differently. He also considers room reverberation part of the spatial image during the performance, and goes a bit further, regarding spectral coloration as part of the interpretation. As a result, although both MOTUS and BEAST use, for instance, speakers facing walls, in order to be able to get 'extra far‑sounding' parts, MOTUS also uses a variety of unorthodox custom speakers to add 'spectral performance' on top of 'spatial performance'.Part of a MOTUS performance. Note the 'ball' and 'column' loudspeakers with their highly coloured frequency responses.Photo: Vincent Laubeuf

NSML systems are not restricted to the playback of pre‑recorded music: acoustic instruments can be integrated into the performance too. As usual, different approaches are possible. Karlheinz Stockhausen's Kontakte has been performed using the BEAST system alongside piano and percussion, and MOTUS sometimes invite instrumental improvisers who react to the spatial imaging that's derived in real time from their performance. At the Groupe de Recherche Musicale in Paris, pieces are also specifically written for the local NSML system, the venerable Acousmonium.

Computer Support

Given the number of loudspeakers involved, NSML system designers naturally resort to software controllers. In Karlsruhe, Germany, the Institute of Music and Acoustics (ZKM) use the Zirkonium control software, mainly to deal with 'super‑panning' across numerous loudspeakers. In Birmingham, the BEAST team use BEASTmulch, a program written by Scott Wilson that manages all aspects of the diffusion system: playback, routing matrices, 'super‑pans', and so on, running in tandem with real‑time human control. BEASTmulch appears to be especially practical in its ability to recall pre‑recorded presets, so that clear and spectacular transitions in the spatial imaging can be made. Use of control software is not universal, though, and the MOTUS team prefer to do without, so that the human performer always remains central to the concert.The BEASTmulch software that controls the BEAST systemPhoto: Scott Wilson

Computer support proves indispensable when using regularly spaced loudspeaker arrays or other special spatialisation techniques. For instance, the ZKM in Karlsruhe uses a 'dome' of loudspeakers that hangs from the ceiling, and BEAST also incorporates a dome‑like array of speakers in the system. Such domes are meant to project the sound in a very specific way, not unlike IRCAM's Spat system, which requires precise volume control that only computers can achieve. Other examples include the use of Wave Field Synthesis techniques, for instance in Parma, Italy, for the Lampadario Sonoro in the Casa del Suono. This special sound‑projection technique dedicated to the creation of virtual acoustic environments naturally finds its place inside NSML systems set up in relatively small spaces.

NSML For Pop Music

As we've seen, for historical reasons, NSML systems are most often used to perform Schaefferian‑style 'acousmatic' music. However, both MOTUS and BEAST have worked with musicians from the 'pop' scene — respectively, Aphex Twin and Scanner — and both Vincent Laubeuf and Jonty Harrison admit that the outcome was not really satisfactory. Apparently, the musicians were primarily concerned about being heard, and there was little time for them to engage with other aspects of the performance, such as spatialisation. This seems surprising, as there is no obvious contradiction between playing loud and taking advantage of more than 70 loudspeakers — quite the opposite, really.

In my opinion, the actual issue here might have been a certain reluctance to lay aside the traditional codes of the stereo image, starting with kick and snare in the middle, along with any lead vocals. It seems likely that proper playback of music pieces including these parts would require them to be on separate tracks so that they can remain immobile, while the spatialisation work would be made on panned material. Obtaining suitable multitrack versions of the material could be a challenge from the technical and copyright points of view. But, that said, imagine a piece like 'A Warm Place' by Nine Inch Nails, played on 80 loudspeakers surrounding the audience. It would make perfect sense in regards to the original intention, especially as far as the 16th-note hard‑panned guitars are concerned. Also, still considering Nine Inch Nails' music, look at the shift in space and arrangement between 3'01 and 3'11 in 'Right Where It Belongs': it almost sounds as if it's specifically written for multi‑loudspeaker systems, reduced to a stereo image for lack of better playback solutions.

To sum up, in my opinion, despite probable initial difficulties to be encountered and solved, intelligent collaboration between NSML system owners and 'pop' musicians would be mutually beneficial to an enormous extent.

There is something extraordinarily refreshing about NSML concerts. Very different from your usual orchestra or band on a stage in front of you, utterly unlike the same two loudspeaker stacks next to the dance floor, and far richer than austere 5.1 or 7.1 surround, properly used NSML systems immerse the audience in sound in a way no other situation can. I sincerely advise every person interested in the sonic aspect of music to go to such a concert at least once in their life. It's a shame that the use of such systems is, for the moment, restricted to a particular kind of music. Let's just hope that NSML system developers will keep on building bridges with other styles of music, and that 'pop' musicians, labels and concert promoters will realise the added value that non‑standard multi‑speaker setups can bring into small and medium‑scale live shows.