High-Performance Computing, Music Composition, and the Soni cation of Scientic Data

Music composition and scientic computing are usually considered separate and un-related elds of intellectual activity. The thought of a musical piece being the result of involved calculations is not a familiar one; nor is that of using sound objects to present the results of large-scale scientic computations. However, an unusual collab-oration centered on the high-performance computing capabilities at Argonne National Laboratory pursues precisely these two ideas. For the past few years, the authors have been engaged in a project to develop high-performance computing tools for the world of sound. Initially, the goal of the project was to see how a composer of serious music might benet from access to the latest supercomputing technology. The idea of using computers for music composition was certainly not new; already in the 1950s, Lejaren Hiller performed many experi-ments at the University of Illinois [1], and the premiere of his Quartet No. 4 for strings \Illiac Suite " [2] in May 1957 is generally regarded as the birth of serious computer music. Computers have helped many composers since to algorithmically synthesize new sounds and produce new pieces for acoustical as well as digital instruments (see Box 1). Given the power of today's high-performance computing architectures, one could reasonably expect a signicant increase in the quality of synthesized sounds and possibly realize a music composition \in real time. " This expectation provided the original impetus for the project