Serial-parallel architecture for real-time synthesis of music
E. Ferretti · The Journal of the Acoustical Society of America · 1979
Parallel architecture is extremely suitable for the synthesis of music. For example the performance of several instruments at the same time is a directly related task for parallel architecture. Serial architecture on the other hand is ideally suited for optimizing real-time tasks. The reasons for serial-parallel architecture are (1) to utilize a single-type module for computing all tasks in real time, (2) to have a means of adjusting the number of tasks without changing the basic system or violating real-time requirements, and (3) to have a means of computing time waveforms which require a large amount of transcendental functions for approximating natural sounds. This paper will discuss the issues of system configuration, accuracy, dynamic range of output versus computation, multiple instruments, and modeling of natural sounds for computer music. Some examples of computer music will be presented, and a high-speed film of a vibrating rubber band will be shown. [This work was supported in part by a private grant from Michael Lay, and use was made of computer facilities in the Computer Science Department.]