The Guardian (Manchester); 20 September 1990; Claire Neesham; p. 33
IMAGINE a concert where the musicians play instru ments that are lines of computer code. These instruments are totally flexible, producing a variety of sounds that is beyond the reach of traditional instruments. Yet the musicians are still performing hitting, sliding or moving objects not merely placing a floppy disc in a drive. The tools needed for this type of electronic music performance are under development now. The International Computer Music Conference, held in Glasgow last week, gave an idea of the type and amount of research being put into playable electronic music. Michael Clarke, a composer based at Huddersfield Polytechnic, says there are three main areas: the production of affordable systems; the development of high-quality, flexible real-time software; and the design of a user interface suitable for composers and musicians. In the UK, the first of these is being tackled by the Composers Desktop Project (CDP) in York. This was set up in 1986 by a team of composers and computer scientists from the University of York to develop an affordable system for computer music composition. The team chose the Atari ST, which was still new in 1986. It was selected for its low-cost, its built-in Midi (Musical instrument digital interface) ports and its powerful programming environment. This was important because a lot of electronic music software is written in languages such as C and Fortran. These programs could be used by the CDP as many are in the public domain or available for a minimal licence fee. One of the first programs the CDP team ``ported'' to the Atari was CMusic from the Computer Audio Research Laboratory at the University of California, San Diego. This provides the basic tools for creating new sound waveforms from scratch. It also gave CDP a springboard for its own software and hardware developments such as the SoundSTreamer, which collects data from a hard disc in bursts and then sends it as sound samples to a PCM AD/DA (analogue-to-digital) converter. The SoundSTreamer supports a SCSI (``scuzzy'') interface as well as connections for digital audio tape (DAT). CDP's software catalogue now includes a signal processing program, Groucho; a Midi toolkit from Carnegie Mellon University; and Barry Vercoe's CSound with FOF (Forme d'Onde Formantique), a time domain formant wave function synthesis program. FOF was written by Xavier Rodet at IRCAM, the international computer music centre in Paris. Clarke went on to add these to CSound. The combination provides composers with a way of producing rich and lively electronic sounds. The CDP is a non-profit making co-operative, so all the software is reasonably priced. The organisation is run by managing director Tom Endrich, who has two part-time assistants and a number of volunteers. The CDP is now looking at ways of speeding-up its system. It can take several hours of processing to produce a few seconds of sound. Clarke points out that if you are waiting days just to hear one small piece of a composition, you tend to lose the creative impulse. There are ways around this. Midi systems, for example, work in real time, are reasonably priced, and have a musician-friendly interface such as a keyboard. But Midi systems are based on packaged processors which can only generate a limited number of sounds. Composers such as Clarke are looking for a system that has the immediacy of Midi but produces the variety and richness of sounds achievable with CSound. This problem is being tackled by a number of groups. Ross Kirk and Richard Orton, at the University of York, are working on a parallel array processor, and the Music Technology Group at Durham University is using Transputers. The Music Technology Group is a collaboration between Durham's music department and the school of engineering and applied science. A system that should provide a real-time environment for composers is being built around 160 T-800 Transputers donated by Inmos. Durham's team has already built systems containing one and three Transputers. Even a single Transputer system was 24 times faster at generating the first 16 bars of Mendelssohn's A Major Organ Sonata than an Atari ST. The three Transputer system is nearly three times as fast again. Peter Manning, Durham's director of music, says it took around one programmer year to convert the CSound code to run on Transputers. However, getting a sound synthesiser to run on the 160 Transputer system could be more difficult. It will involve writing a sound synthesis system from scratch, using a specialised programming language such as Occam. This task may appear daunting but Peter Bowler, a member of the group, is confident the results will be worth the effort. ``The purpose of the system is to enable us to make the leap to real-time'', he says. The system will still understand the CSound language, which requires the composer to write two programs one which specifies the instruments and one which codes the orchestration of the piece. However, this type of composition is not for the number-shy. Therefore the Durham team will be looking at new ways of interacting with the system. At present the electro acoustic composer often has to do everything beforehand so that the ``performance'' consists only of pressing the play button on a tape recorder. The long processing times makes live performances impractical. A flexible real-time system could make the electro acoustic composer's life simpler. It also could help promote the type of electronic music that goes beyond Midi rock and pop. Rob Watson, one of the composers at Durham, points out: ``Once real-time systems become stable and cheap, it will be one stage to increasing the accessibility of electronic music.''