An audio data format in which an instrument is described using a combination of sound samples and articulation instructions which determine modifications made to the sound sample is provided. The instruments form a first, initial layer, with a second layer having presets which can user defined to provide additional articulation instructions which can modify the articulation instructions at the instrument level. The articulation instructions are specified using various parameters. The present invention provides a format in which all of the parameters are specified in units which relate to a physical phenomena, and thus are not tied to any particular machine for creating or playing the audio samples. The articulation parameters include generators and modulators, which provide a connection between a real-time signal and a generator. The parameter units are specified in perceptually additive units, to make the data portable and easily edited. New units are defined to give perceptual additive parameters throughout.