Behaviour, Structure and Causality in Procedural Audio1

Behaviour, Structure and Causality in Procedural Audio1

Andy Farnell (Computer scientist, United Kingdom)
DOI: 10.4018/978-1-61692-828-5.ch015
OnDemand PDF Download:


This chapter expands some key concepts and problems in the emerging field of procedural audio. In addition to historical, philosophical, commercial, and technological themes, it examines why procedural audio differs from earlier “computer music” and “computer sound”. In particular, the extension of sound synthesis to the general case of ordinary, everyday objects in a virtual world, and the requirements for interactivity and real-time computation are examined.
Chapter Preview


Procedural audio is sound as a process. Instead of thinking about nouns, we think of verbs, or the actions that cause sounds. Procedural sound is also a structural and reasoned, rather than purely sensible, approach to sound, in which behaviour supplants identity and we ask not what sound is, but what it does.

This chapter follows the publication of Designing Sound (Farnell, 2008) in which I lay foundations for procedural audio, particularly its use in real-time virtual worlds. Here I would like to talk about the ideas of behaviour, causality and structure which are implied by the identification of sound as a branch of dynamics requiring energetic change.

The task at hand is producing sound for film, computer games, or other interactive entertainment applications. This task requires creativity, knowledge and understanding. Creativity in traditional sound design is directed at capturing, curating and matching audio data to depicted circumstance, whereas creation of procedural sound is from first principles: so it is truly design as opposed to selection. Insights into process are therefore at the root of the work.

The end product is sound as code, or sound objects (Polotti, Papetti, Rocchesso, & Delle, 2001). One must create new digital signal processing (DSP) algorithms, or a set of parameters for existing sound objects, rather than actual sound files. The product is “potential sound”, rather than particular audio data as it would appear at the digital analog convertor (DAC). The complete package of a sound object is DSP, control, and encapsulation code compatible with a set of parameters to be supplied in the future. Objects may be instantiated and animated at some later time, in any contrived circumstance. We say this media has deferred form, exhibiting desired behaviour according to supplied runtime parameters.

Complete Chapter List

Search this Book: