The Music of CSIRAC: Australia's First Computer Music (review)

Readers of Computer Music Journal will likely be aware of Paul Doornbusch's work on early computer music in Australia (see his Spring 2004 article, "Computer Sound Synthesis in 1951: The Music of CSIRAC," 28(1): 10–25). For many, it was a revelation to realize that music was being produced "down under" as early as 1951, several years before Max Mathews and his colleagues were producing digital sounds at Bell Labs in New Jersey (but at just about the same time that the Ferranti Mark I computer in the UK was also producing simple musical examples). The Music of CSIRAC: Australia's First Computer Music presents a fuller account of Mr. Doornbusch's research into the music-making capabilities of the CSIRAC computer in Sydney and Melbourne and his efforts to reconstruct the sounds that had been produced there more than 50 years ago. Although a few of the recorded sound examples were included on the CMJ DVD Vol. 28, this publication comes with a CD-ROM containing a larger collection of the music produced on the CSIRAC.

As Paul Berg points out in his introduction to the book, "the path that led to [computer music] is the result of many very small steps. Steps that were neither coordinated nor goal-oriented. Steps that were not labeled merely scientific, consumer-oriented, or artistic" (p. ix). At Bell Labs they were researching the transmission of speech signals for telecommunications systems; Lejaren Hiller was a researcher in chemistry with access to the ILLIAC computer at the University of Illinois, Urbana Champaign; Iannis Xenakis was trained as a civil engineer. There was no research agenda for developing the field of computer music, and little direct communication between the people carrying out pioneering work in different places (between Illinois and New Jersey, let alone between Australia and the USA). In Australia, the innovative work on creating programs to produce music was done by mathematicians, physicists, and engineers.

The Council for Scientific and Industrial Research (CSIR) developed one of the world's first stored-program electronic digital computers. The CSIR Mk1 ran its first program in November 1949. In 1955, the computer was moved to the University of Melbourne, renamed the CSIRAC, and held in service until 1964. The computer has been preserved relatively intact (although not operational) at the Museum Victoria, where Mr. Doornbusch was able to access it (along with much related materials, such as program tapes, documentation, etc.) in order to reconstruct the historical procedures and technology developed by the early Australian researchers for producing sound. One of his aims was to reproduce the instructions for the music programs through the same process as would have been carried out originally, output through the same loudspeaker driver that was installed as part of the computer. This work involved much effort trying to decode the sometimes ad hoc instructions, and trying to understand how the programs were run and how the sounds were created.

The CSIRAC did not contain a digital-to-analog converter; such technology was nonexistent when the computer was built in 1948– 1949. Instead, program instructions could be routed to the loudspeaker on the data bus as a series of pulses, channeled through an amplifier. Initially, the loudspeaker was intended to provide an audible signal that the computer was functioning correctly (or incorrectly). But, it quickly became apparent that by sending multiple pulse instructions or by looping them a continuous tone could be produced. From there, the engineers worked to develop methods for generating different frequencies, and by extension, melodies. The computer was never able to produce more than one pitch at a time, and the waveform was fixed, as a...