: ZKM :: Institute und Abteilungen :: Institut für Musik und Akustik :: ZKMusik :: veroeffentlicht :: PD_Futuristic

A futuristic audio network
Center for Art and Media Technology Karlsruhe

Dr. Ing. Pierre Dutilleux

Center for Art and Media Technology
Institute for Music and Acoustics
Karlsruhe, Germany

1. Abstract

A building-wide digital-audio network is outlined for a cultural center that will include 3 control rooms, 6 studios and performance hall, many work places for audio production. When most of the equipments are digital, the suitable network is digital.

The combination of the technologies of digital audio, computer networks and fiber optics are envisaged to meet the goals of the center.

2. Introduction

The purpose of this talk is to introduce the center and to bring ideas about how could the audio network be realised, in a building were most equipments are computers or multimedia terminals.

I will begin with the presentation of the Center for Art and Media Technology, describe briefly the future building and give examples of its current activities and productions.

The audio network will provide many various services. I will comment the ones that are more specific to our domain. An idea needs a body to become a reality. This body will be provided by the technology. We can choose between several ones. I will review some key ideas of the choice of a technology.

3. The Center for Art and Media Technology

"The Center for Art and Media Technology Karlsruhe (ZKM) brings the arts and new forms of media together in theory as well as in practice. ZKM aims at fostering the creative possibilities of a connection between traditional forms of art and new media technologies, to achieve results which anticipate the future."[]

The center is comprised of the Museum of Contemporary Art, The Media Museum, the Institute for Image Media and the Institute for Music and Acoustics. The center is in the process of building up. It will be fully operational by 1996, when the new building is finished. For a couple of years, it was agreed that the future building would be that one designed by the Dutch architekt Rem Koolhaas, but the local authorities have recently made the decision to move the center to another building. The center will be housed in a large historical building of the city, the IWKA. Compared to the design from Koolhaas, it should offer as much or even more possibilities.

Aside from the collection and presentation of art works, the center places its emphasis on the production of artistic works in the various institutes. Though the main tool for artistic production is the computer, digital technology does not necessarily have to be part of all works. "The core of ZKM is found in the computer laboratories. These are the places for experimentation and research, where artists and scholars have the necessary equipment at their disposal to apply new technology to aesthetic practice."

The Institute for Music and Acoustics is open for artistic and scientific work in general areas of production of musical works (tape, live-electronics, sound installations with or without other media), open processes for learning and discussion (reflection of esthetic conditions), the development of computer music environments and systems, scientific research (psychoacoustics, basic musical research). Multimedia projects can be produced through the collaboration with the Institute for Image Media. For the time being, the Institute for Image Media focuses its efforts on computer animation, the construction of interactive installations and architectural simulations.

"Bringing the work produced in the laboratories to the public and exposing this work to a critical public eye necessitates special performance facilities which do not exist in conventional performance spaces." A large performance space (900 m2), the Media Theater, will serve this purpose.

One must know that we are already effectively working, eventhough we don’t have our own building. We currently work in three places in the city of Karlsruhe. One place for the administration, on for the Instistute for music and acoustics, one for the Institute for bildmedia and Museum of media. Waiting for a place to be shown, the art-works of the museum for contemporary art are stored in large warehouses. We organize concerts, multimedia performances and exhibitions. We release publications. We support artists with scholarships, technical equipment, assistance from technically qualified people and, to a very limited extent, with housing.

4. Some "audio-operations" at ZKM

The activities in the full scale ZKM, when the building is finished, will be manifold, but let us consider some typical operations that are performed currently in the Institute for music and acoustics. For these operations, we wish we had special machines. For economical reasons, we buy standard machines from the audio industry, although some of our requirements are not fullfilled by these machines.

4.1. Sound generation

The issue of sound generation is manifold. It is available to the public through synthesizers. Musicians can choose within a very broad range of sophisticated machines. Though, the universal machine does not yet exist, so, many composers design their own sounds with various programs running on standard computers. At ZKM, a special program is used for this purpose : Common Music, it is the fruit of a collaboration with the Center for Computer Research on Music and Acoustics (CCRMA, Stanford). Some composers need real time systems, some do not, but most of them wish the computer to become faster and smaller. To answer this wish, the Institut de Recherche et de Coordination Acoustique Musique (IRCAM, Paris) has developed a digital signal processing board and a set of musical softwares. This board enables the composer to design interactively algorithms for sound-synthesis and -processing. Thanks to the high speed processors, complex algorithms can be used in real time.

4.2. CD production

The Compact Disk (CD) technology offers a good quality for the final production of stereo signals. With the advent of WORM disks that can be written to meet the CD standards, the medium becomes an attractive alternative to the tape. Although the medium and the recording equipment is still expensive, it is affordable for a limited number of copies.

The main drawback of the CD format is its stereo format. Since the beginning of electroacoustic music, the composers are eager to produce multichannel pieces (4 or 8 channels are frequent). The flexible and reliable medium that will handle this format carefully is still awaited.

4.3. Multimedia and archiving

Multimedia applications is growing on the flexibility of the audio sources. The remote control of digital equipments and the access to large audio data bases is necessary.

The students in musicology or composition are often talked about interesting electroacoustic music pieces, but they seldom can access and listen to them. There are many reasons for that, I will quote two of them. The one is that the work was recorded on a medium that is degrading with the time, the other is that no copy of the work is available. We hope that the digital technology will help to restrore old works and, thanks to its flexibility, will enable the survival of the copies through the future improvements of the technology. Large audio data bases, available on CDs or on a network, will help disseminate the knowledge about electroacoustic music.

5. The goals of the audio network

5.1. The places to interconnect

The new building will offer many facilities to each of the departments of the center. Most of these facilities will be technically interconnected in order to promote interdisciplinary work. Let us consider the places that are concerned by the audio network. The area for music production will include one large recording studio (250 m2), with the main control room ; four smaller studios for musical work and scientific set-ups (40 m2 each), sharing a control room ; five rehearsal studios for composers and instrumentalists as well as offices for guests. Aside from these places that have an emphasis on music, many other places througout the building are related to music. The Media Theater and its audio control room have very strong links to the Institute for Music and Acoustics. Each of the Museums, the libraries for audio and video, the conference and show rooms, the roof (for open air concerts), several offices will have access to the audio network. All in all, the number of 40 rooms to be interconnected is reached (Figure 1).

5.2. Versatility

The center will promote artistic projects of any size. One can think of several small projects that time-share the ressources of the recording studios or of the audio library. On the other hand, there could be a gigantic project that involve simultaneously all the ressources of the building. Within this frame, the building should be considered as a flexible machine, each fonctional part of which could come under the control of an artist or of a performer.

As an example, let us consider a musical performance that involves live performers, live-electronics and recorded sound elements (Figure 2). This set-up is typical of the large performances such as those performed by IRCAM or Experimental Studio Freiburg. An orchester is located at the center of the theater. The public is sitting around the orchester. Between the orchester and the public are a few solo instruments. Some sounds, prepared in the studio, are played through loudspeakers. The solo instruments and the orchester are picked up by microphones, processed by a real-time computer and played back through loudspeakers. The sound processing comprises timbre alteration, and spatialisation. In order to achieve consistent spatialisation effects over a wide area, many loudspeakers are set up around the orchester. Some parameters of the musical composition are specific to the electroacoustic installation. They specify the modifications of operational parameters againt the time. These modifications are automated thanks to control computers. The electroacoustic equipment is based on a large mixing console, a multitrack recorder, real time computers for sound processing and a matrix to produce spatialisation effects. The system is under the control of a computer that schedules the parameter updates. The multitrack recorder delivers the sounds that were prepared in studio. The matrix has typically 24 inputs and 24 outputs, it receives processed signals from the mixing console and sends to each loudspeaker a special combination of these sounds. At each point of the matrix is an element to modify the amplitude of the signal (a Voltage Controled Amplifier (VCA)).

Such a performance involves so many people and equipments that it can seldom be renewed. In order to make the piece of music available to a larger audience than that of the concert, stereo or 8-track versions will be produced from the recording of the concert. For that purpose, during the concert, the multitrackrecorder is stetup in order to record some of the audio signals that are going around the system (sounds from the solo instruments, sounds that have been processed in real-time, ambience...).

The set-up that has just been described should not be hardwired in the building, but it should be possible to configure the equipments and the studios in order to realise it.

5.3. Quality

To be accepted from broadcasting institutions or simply from the public, the productions from the center should meet high technical quality standards. At the time of digital audio, one could simply say : "let’s go digital". It would be fine for simple equipments, but in the professional domain, the all-digital (big) studio ist still a challenge. So, we will consider a mixed audio network, with analog and digital links.

5.4. Reliability

The audio network should be reliable. That is one of the main qualities of technical equipment, but flexibility is often traded for reliability. May we take the buildings of the broadcasting corporations as an example ? The broadcasters have high reliability standards. Protections are built in the equipments to prevent hazards such as line break-up, overmodulation, or to restrict the access to the routing and scheduling devices. To find a compromise, we should consider the various activities that are going on in our building. We can see three categories of activities : performances and concerts ; production and on-line composition (the composer is working at a computer that processes sound interactively) ; research and off-line composition (the composer is working at a computer with abstract tools, no sound processing is required at this time). The performances and the concerts may not suffer from a line break-up, they will demand a high reliability during a given period of time ; the production of pieces and the on-line composition need to be allocated some ressources during time-slots, if a technical problem occurs, the process can usually be rescheduled ; the research and the off-line composition need to know when and how a projekt can be realised, but during most of the time, the work can be pursued in a virtual world. For these last activities, the flexibility of the equipment is of paramount importance.

So, we should consider an audio network with at least two levels of reliability. A high reliability network, for performances and concerts, and a highly flexible network for the other activities. Both network-levels could share the same pool of equipments, but the subsets should be safely separeted by using different physical connections and control-equipments.

6. Technology

A recording studio was traditionally equiped with analog electronics. High performance cabling standards have been developed to serve these equipments. Today, we use many more digital equipments than analog ones. It could be wise to consider a technology that is closer to the computer technology than to the analog electronic. Furthermore, the fiber optic technology promises to open highways to our digital sounds. Let us briefly review some of these points.

6.1. Traditional solutions

Until a few years ago, the recording technology has grown along with the analog technology. Nowadays, the digital technology is superseding step by step each element of the studios, but, as far as I know, the largest audio installations still rely on analog technology. Using this technology is still the simplest way to ensure that all the equipments can work together without facing a nightmare of format conversion and of synchronisation.

In our future building, in order to interconnect all the studios and rooms above quoted, we initially planed a topology with many star connections and a few ring lines. We were in need of a 800 x 500 patchbay. Taking into account the fact that not all lines will be simultaneously in use, the size is now estimated to be about 150 x 150. We intend to have additionnaly 4 ring-lines going through each of the rooms. From a one-level network, with a single switching element at the center (Figure 1), we evolve towards a multilevel network, with several small switching elements at the periphery of the main switching element (Figure 3).

6.2. Computer networks

The computer could very well edit text files, assist at off-line composing. It now manages worldwide electronic-mail, from one’s desktop it is possible to work on a remote computer. Different machines are connected with lightweight cables, the processing power has grown to become usefull for interactive composing. One thinks about new applications and about traditional activities that can be done in another way. The temptation is high to believe that the audio network could run on a computer network. Theoretically, all what we wish can be realised, but at which cost, financial and human ? Let us reconsider the traditional working methods in the studio, the opportunities of LANs and WANs, and the bandwith and topologies of nowadays computer networks.

6.2.1. An integrated studio

Now, let us have a systematic approach to the description of a studio. Let us consider it as a system with inputs and outputs (Figure 4). The system must switch many lines, mix many inputs to produce many outputs, process several signals (filtering, amplitude compression/expansion, reverberation...). The outputs are loudspeakers, tape recorders, auxiliaries. The processing function can be carried out with independant subsystems. The switching and mixing function can be realised within a unique matrix where each element is an amplitude modifier. The tape recorder can be seen as a mass storage device. Some companies have put this concept at work. It turns out that the number of signals and the amount of processing that are involved is very high. No single-processor computer can handle the whole task, that means that multiprocessor architectures are compulsory[].

6.2.2. Local Area Networks

We are currently using computer networks for electronic-mail, data bases, file exchange etc. Most of the traffic can accomodate a fairly low throughput, and the computer networks have been designed for these applications.

In the future, we intend to make a more intensive use of computer networks, to provide access to multi-media data bases, and to have distributed performances. For example, the Institute for Image Media is experimenting the concept of graphic art on the network. Several artists, located in different cities, could work together, interactively. The medium here is ISDN. The advantage of ISDN is that it is planned to be as wide as nowadays telephone networks, but we expect to be able to use other networks. To meet our quality standards for sound or picture, we need larger bandwidth than that offered by ISDN-S0.

6.2.3. Digital audio on the network ?

I will compare the requirements of digital audio to the properties of the usual computer networks[,]. A stereo digital-audio signal has a bit rate of about 3 Mbit/s. Can we consider transmitting such a signal using popular computer networks ? Let us consider the LANs such as LocalTalk, Ethernet, Token Ring and FDDI. In the table are listed the properties of these networks that are relevant here.


Speed (Mbps)

Access mode


Access time











Token Ring

4 or 16

Token passing





Token passing

Dual ring


Because of ist low speed, LocalTalk cannot be used ; because of its unspecified access time, Ethernet cannot be used in real time, but thanks to its speed, Ethernet can be used effectively to transfert sound files ; if the number of applications sharing the medium is very limited, real time operation could be achieved, but with no guaranty. The Token Ring network features a high speed and a specified access time, this network could effectively be used for an application with a very limited number of channels. The single ring structure of Token Ring provides no safety in the case of a station going into troubles. On the other hand, FDDI features a dual ring. This ensures that should a station go into troubles and break up the transmission over the primary ring, the transmission could recover by using the secondary ring. With its specified access time and high speed, it could theoretically accomodate in the order of 25 stereo channels of digital audio. Such a capacity can be usefull to interconnect all the places in our ZKM building that need access to a few signals. But it is still too limited to realise the whole audio network. FDDI, using optical fiber, is far from exhausting the transmitting capabilities of its medium, so we are confident that, in the future, faster standards will become available. The effective transfert rate that can be achieved on the network is, again, very dependant on the access method that is implemented[].

Networking means not only transporting but also picking and delivering, I mean that the addressing system is a key element. In the traditional audio networking, each line carries a single signal, source addresses and destination addresses are identical to the source plug and the destination socket. This is one of the reasons why the switching matrix reaches so big a size in our project. In computer networks, source and destination address are numbers that remain associated to the signal that is carried over the medium. The source of digital audio signal (the transmitter) transmits his address number along with the signal. The device that receives the audio signal (the receiver) has a number as destination address. It scans the medium for this number, when the destination address of the signal matches the number of the receiver, the receiver knows that the signal is for himself. The receiver then copies the signal from the medium and the signal becomes available for further audio processing.

6.2.4. In practice

Let us come back to practice. I do not know today of any system that applies truly FDDI to audio networking, but there are a few applications of optical fiber at a similar speed and using the same optical fiber as a medium. The first to be standardized is the Multichannel Audio Digital Interface (MADI). It is a point to point connection for 56 digital audio channels. It finds its first application with the interconnection of multitrack recorders and mixing consoles. Other applications of optical fiber for distributed sound transmitters/receivers are available or reported, but as stand alone systems. That means that it could be questionable to build an all-purpose audio network around these systems[].

6.2.5. Control of audio systems

I have been dealing here with the sound signal itself, but the control signals could require another discussion. The number of equipments in an electroacoustic installation and the capabilities are increasing, so that remote control systems become necessary. The AES standards comittee is working on this topic. We hope that this standization effort will bring attractive systems on the market.

6.3. Fiber optics

The optical fibers give us a communication medium that has a very large bandwidth. In order to accomodate such a bandwidth, special devices are necessary. One needs ransmitters and receivers to convert the signal into light and vice versa, modulators to drive the transmitters from the audio signal, demodulators to recover the audio signal form the light receiver. To connect the differents transmitters to the diferent receivers, special commutation techniques are available. In all of these areas, the technology is progressing fast and new products are in development.

The processing of optical signals is promising, although the fonctions are still limited. The optical switches, 1 to 2 for example, are readily available. The control signal is a simple TTL level. If several tranceivers have to be connected together via optical fibers, it is practicable to make a star connection, the fiber being interconnected at the center. The interconnection of an optical fiber to another is succesfull if the light from the first gets into the second. This can be realised by twisting the fibers together. For this application, the cladding of the fiber has to be removed before twisting.

The bandwidth that is made available by a single optical fiber is so large that it can be shared by several users. The analogue electronics technology was used at frequency division multiplex (FDM). The digital electronics technology has replaced FDM by time division multiplex (TDM). In fiber optics, implementing TDM means converting the optical signal to electronics, multiplexing in the electronic domain and converting back to light. The electronics that is involved is high speed (ECL). To avoid using so much of high speed electronics, it is possible to multiplex the informations on the fiber using different wavelengths. We have here a wavelength division multiplex (WDM), which could be compared to the earlier FDM[]. The advantage of the WDM is that it reduces the use of high speed electronics. Its drawback is that it demands high accuracy and stability from the light sources and the light receivers.

The national telecommunication administrations, or companies, are developing services that will make use of the bandwidth of the optical fibers, and that will go to the home, as the telephone does today. To reach this goal, integrated transceiver modules are developed. They feature both low cost (when compared to previous designs) and high performance, as to accuracy and stability of the operating wavelength [].

7. Conclusion

At the Center for Art and Media Technology, we aim to build an audio network that is up to the technologies that are used througout the Center and that has the qualities of the traditional analog cabling.

We could see our network as a line-switching-network, each line being medium bandwidth (1Mhz), and all the lines being sampl synchronous. This last characteristic seems to be rather peculiar in the realm of the multimedia world that is in development by the telecommunications industry.

The optical fiber has the advantage of its high speed, but the analog cabling has the advantage of a simple interface between equipments.

If we consider the topology of a Wide Area Network as a template, could we design an audio network comprised of a global line switching system and of a few dedicated lines for the main consumers. The currently available fiber optic systems (100 Mbps) handle about 50 channels of digital audio. If we estimate that we need in the order of 300 simultaneously active connections, we would need several such systems to be interconnected. Could these connections be realized through a high speed backbone (1 Gbps) ? The dedicated lines could be realized using MADI connections and a few optical switches (Figure 5).

In the coming years, we will have to decide according to the possibilities of technology, but we will keep in mind that the system should remain useable and serviceable by audio engineers, without demanding from them to become computer network specialists.


8. References

9. Figures

Figure 2 : A concert with live electronics.


Figure 3 : A structured network.


Figure 4 : The studio, seen as a system.



Figure 5 : An optical fiber-based network.


© 2014 ZKM | Zentrum für Kunst und Medientechnologie Karlsruhe :: Impressum/Web Site Credits