With procedural audio, instead of playing recorded sounds, we define mathematical models that represent these sounds. The models can then be used to generate audio signals, based on the values of their parameters, and real-time inputs from a designer, a game engine, an animation curve etc...