would not be modified, only the order of the sounds, or the measures, or any time unit that was chosen for the shuffling. When time units became extremely short (ranging from 1 to 40 milliseconds) the transformation affected principally the spectra creating very interesting granulation processes.

SYTER was the next step; the project evolved towards a system capable of doing real-time sound transformation, synthesis, hard-disk reading and writing, hard-disk editing; and all in a very easy-to-use graphic environment. The project in its actual look was finalized in 1985 and it involves a signal processor, a host computer (PDP 11/73), a hard-disk, and most important; a software environment, sufficiently supple to be used as a closed system with various prepared configurations; and as an open system permitting module programming to construct new transformation objects.

Both the hardware and the software were developed by Jean-François Allouis and Jean-Yves Bernier, in a first period, and finished by Hugues Vinet. The system was commercialized by a french enterprise and 10 SYTER systems were built and sold (one must keep in mind the enormous cost of these systems, ranging from 100,000 to 150,000 US$ depending on the options).

From 1985 to 1990, moment in which the developments on this system were stopped, the essential research on the system consisted in algorithm improvement and development. Essentially all the software existing on the 123 programs was transplanted to SYTER, sometimes with certain limitations due to the necessity of keeping up with real-time. About 50 different 'instruments' (name given to the finished algorithms) were developed and different users made about ten times more instruments consisting of derivations of the first group of 50. Two SYTER systems are still working in our studios.

GRM Tools

Once the SYTER project was finished in its principal aspects, the next step was to try to adapt the transformation algorithms to personal systems. Technically this was a very difficult task, since personal systems were far to slow to guarantee a real-time perception (real-time is real only for our perception, since this implies that calculations on sound are done so quickly that we perceive the input and the output as being simultaneous, in fact a small delay always exists), and this was only possible with the arrival of the Sound Tools system by Digidesign, that used the 56001 DSP's.

First a programming environment was developed in the system by Hugues Vinet, called DSP Station and a certain number of the SYTER algorithms were transported on this Station. The result was GRM Tools, a package of ready to use, graphic controlled, real-time algorithms.