High Level Musical Control of Sound Synthesis in OpenMusic

̊ new classes have been added to OpenMusic in order to handle the big amount of parameters involved in the control of synthesis. OpenMusic (Assayag & al, 1999) is a computer assisted composition software based on Common Lisp / CLOS (Steele 1990). It is is a complete visual programming interface to CL/CLOS where user operations are mainly based on the drag and drop scheme applied over icons, which stand for any OpenMusicrepresentable objects, and containers, which are editable panels giving access to the internal structure of objects. As CLOS is metaclass language, classes are a particular kind of objects, which are in that case instances of metaclasses. So classes may be handled visually as icons, and their structure built or edited through their graphical containers. The same stands for generic functions, which containers own a set of methods. OpenMusic being a musically-oriented environment, a lot of predefined classes and methods are made available to handle musical structures and allow their graphical edition (including, of course music notation). After a lot of investigations in the field of symbolic computation, we have decided to explore the potential of the OpenMusic visual language as a control environment for sound synthesis. The idea was not to implement synthesis engines into OM, but to handle visually different high level control paradigms, then to generate the low level parameters towards whatever engine was available around. We have set up a collaboration with Marco Stroppa who has been involved for a long time in the control question and who had set up his own portable environment, called Chroma, in pure Common Lisp. The question adressed by Stroppa was in particular how to manage huge quantities of control data, bringing them back to a few, musically significant abstractions. He had designed a matricial representation, where rows and columns would set up vectors of values for identified synthesis parameters. The good thing was that values could be given as litterals as well as functions, thus providing the abstraction needed. Another advantage was that these functions could be easily connected to compositional processes, such as harmonic or rhythmical ones, independently from the constraints of synthesis. As OpenMusic provided already a powerful framework for composition, it seemed natural to implement this matrix abstraction in order to articulate symbolic composition and sound synthesis. It was also a challenge ̊: would the visual object oriented environment resist and prove ergonomy when confronted to the large sets of data involved by synthesis ̊?