Procedural audio synthesises sounds algorithmically from parameters rather than replaying recordings

Procedural audio (also called model-based or sound-design synthesis) generates sound in real time from a functional description rather than playing back a pre-recorded sample. The approach is to reverse-engineer the physical or perceptual structure of a target sound and rebuild it from basic DSP blocks (noise sources, filters, oscillators, envelopes). Because parameters drive the synthesis, the result responds continuously to game state, performance input, or generative processes: the same patch can sound like a distant campfire or a roaring bonfire by changing a single control signal. The trade-off is that procedural audio requires DSP understanding and more design time than sample playback, but exposes controls (intensity, size, material) that a fixed recording cannot offer.

Examples

Farnell’s fire patch (practical 11): noise filtered into crackle, hiss, and lapping layers — no sample file needed. Rain could similarly be built from randomly triggered impulse noise shaped with a short decay, with a control for drop density.

Assessment

Name two advantages and one disadvantage of procedural audio over sample playback. Give an example of a sound property that would be difficult to control in a sample but trivial to expose as a parameter in a procedural patch.