In this article, the author argues that there can be several levels of musical data representation and solutions to musician-computer communication problems.
1. On the first level the composer adapts her/himself to the machine working system for sound production. This is the operational score (the definition of each sound event depends on physical models for simulation and production, and on specific physical parameters; one type is the MUSIC score, which is divided between the digital instrument, simulation algorithm, and the temporal definitions that control it). 2. On the second level the composer uses symbolic language for codifying musical information in relation to a perceptive psychological model or a conventional model defined by her/himself. This is the symbolic score, one that is based on perceptual dimensions; software must pass through the interpretation of symbols and its translation into the machine language. 3. On the third level the composer stresses formal models of musical composition; defining, organizing and operating on sets of musical parameters or sound events according to a more general plan. This is the structural score which refers to the formal model of traditional instrumental composition; it permits the definition, organisation and elaboration of parameters both on a macro and micro scale; it can be chronologically-associative, top-down, bottom-up, in blocks; it is from time to time operational (e.g. PLF programs in MUSIC software); symbolic, based on perceptual models (e.g. Projekt by G.M.Koenig) and mixed if it elaborates operational and symbolic parameters (e.g. EMUS software realized at CSC (Centro di Sonologia Computazionale dell'Università di Padova)).
In general, the article concerns the analysis of various levels and their interaction. Special attention is paid to the interpretation of the symbolical data and to the results of the work with EMUS, a program for musical structure processing realized at CSC.