Music Representation

Common Music is designed to support both algorithmic and non-algorithmic styles of music composition. In algorithmic music a composer specifies musical material indirectly through a program that creates it. In non-algorithmic music a composer specifies musical material directly by defining it. Algorithmic music might be termed a procedural style of music composition in which music is represented as a set of instructions. Non-algorithmic music might be termed a literary style of composition in which musical symbols are created, edited and transformed until a composition is finally completed. Composers often want to mix or combine algorithmic and non-algorithmic style together.

A basic motivation behind Common Music is to provide a music representation system that permits both literary and algorithmic styles of composition to exist simultaneously and to expand algorithmic processing to include editing operations as well as generative ones. Since textual and algorithmic material may be processed together simultaneously and in real time, a composer is completely free to choose the model most suitable for describing a particular portion of the work in progress. A composer may also choose what not to process, either by defining alternative layouts of the same structural material or by temporally marking events or sections as 'hidden'. A hidden object is skipped during musical processing.

The primary distinction in Stella's representational scheme is between classes of objects which in some sense reflect basic compositional data and classes of objects which implement collections, or groupings, of other objects. This basic distinction is represented by two early classes in Stella: Event and Collection. An Event represents an atomic structure that cannot group other objects. This does not mean that an event does not contain further subdivisions - for example, into slots or parameterized sound data - but only that the system's mapping, editing and selection functions operate on the event as a basic structural unit. The Event class is the root class for specializations of musical data meant to be output to some sound synthesis program like the Music Kit, CLM, CSound or MIDI. Events have no notion of absolute time, but only of time increment. Since Events maintain only a time increment to the next object, they may be copied and pasted to different locations in a composition and may be structure-shared by different Collections.

In contrast to an Event, a Collection represents a grouping, or aggregate, of zero or more sub-objects. Sub-objects stored in a Collection may themselves be Events or other Collections. In order to represent aggregate compositional structure, several different types of collection are necessary. Each type represents a particular order in which Events in the Collection should be accessed or processed in the synthesis layer.

There are currently five broad subclasses Collection: Thread, Heap, Merge, Algorithm and Generator.