Microsound

Curtis Roads

Sonogram made from a microsound recording on the enclosed CD. The composition is made up of many sound objects, which appear as star-shaped figures.

lec­tron­ically-generated sound is a big business these days. How­ever, early composers of electronic music like Iannis Xenakis and Edgard Varèse found their works stuck on the sidelines as electronic music came to be associated with low-budget sci-fi flicks and schmaltzy imitations of conventional instruments. Part of this was probably due to the limitations of early instruments like the theremin and Moog synthesizer, which were crude and inflexible by today's standards. Computers are now capable of synthesizing any waveform that is desired, which means the complexity of the sound they produce now rivals that of acoustic instruments.

Curtis Roads calls such computer-generated sound "microsound", by which he means sound that is created by digitally piecing together waveforms from building blocks or "particles" whose individual length is between 50 and 100 milliseconds. Composers using this technique face a formidable challenge. They must not only create a meaningful composition, but also visualize the basic sound for an instrument that is programmed instead of played. Their audience may even be hostile, regarding pre-programmed sound as little different from a recording.

An example of microsound is shown in the sonogram at right, which I made from one of the audio samples on the CD in the book. In this composition, one single waveform was repeated many times at different frequencies to create a cloud of sound. Working with small chunks is much easier than the old method of specifying the complete waveform in the voltage domain. Using that technique, at the standard CD rate of 44,100 samples per second, the composer would die of old age after producing only a few seconds of music.

In this book, Roads presents some psychoacoustics and elementary signal processing theory, mixed with descriptive language intended to inspire musicians to try out his ideas. Although the book is too basic to interest engineers, musicians may be inspired by it. They're already used to working in the Fourier domain, so it's a small step to manipulating frequency spectra to create sound. But Roads still has to spend so much time explaining basic signal processing concepts that there's little time left for building on his ideas about grains of sound. How do you get grains of sound to interact to produce music instead of noise? Roads devotes the last two chapters to this issue, but doesn't get very far.

The computer opens up a new universe for composers, who can now sculpt with raw sound. Maybe it shouldn't even be called music, but a new art form based on sounds. This book will give the basic concepts needed to make those sounds. But in the end, it's up to the composer to figure out how to turn raw sounds into a language, so the computer does more than just make cool noises.