ï~~
An Introduction
to the
SSSP Digital Synthesizer
William Buxton, E. A. Fogels, Guy Fedorkow, Lawrence Sasaki, and K. C. Smith
The Structured Sound Synthesis Project
Computer Systems Research Group
University of Toronto
Toronto, Ontario
Canada
MSS 1A4
1. Introduction
One of the main interests of the Structured Sound
Synthesis Project (SSSP) is to develop a highly interactive
environment to serve as an aid in the composition of music.
We would like to do this within the context of a small,
inexpensive, accessible system. In order to resolve the conflict
between the high computational demands of sound synthesis
and these design objectives, we have developed a specialpurpose digital sound synthesizer. Work on this project began
in January 1977, and the device was functioning in a limited
state by December of that year.
The design owes much to the Dartmouth synthesizer
(Alonso et al., 1976) as well as the VOSIM oscillator (Kaegi
and Tempelaars, 1978). Generally described, the synthesizer
consists of one "real" oscillator which is time-divisionmultiplexed so as to function as sixteen oscillators. There are
two aspects of the device which we see as particularly significant. First it was designed so as to incorporate in hardware five
important techniques of sound synthesis: fixed waveform,
frequency modulation, additive synthesis, waveshaping, and
VOSIM. Secondly, the device can be easily interfaced to most
digital computers.
The generators are essentially fixed sampling rate,
accumulator-type digital oscillators. The sampling rate of each
oscillator is 50 kHz, with a system bandwidth of 20 kHz.
Dynamic range is well over 60 dB (and should improve with
further adjustment) while frequency resolution is approximately.7 Hz (linear scale) over the entire bandwidth. This will
shortly be improved to enable resolution of less than one
"cent" at even extremely low frequencies'. The output signal
of each oscillator can be fed to one of four analogue output
busses which may then either be fed directly to an amplifier,
or to a channel distributor (Fedorkow, 1978; Fedorkow,
Buxton and Smith, 1978). The waveform output by each
generator may not only be user defined-up to eight wave
forms available at one time-but one may switch waveforms
in mid -cycle. This is possible since the sixteen oscillators
share a 16k buffer of 12-bit words to store waveforms. This
16k of RAM is partitioned into eight 2k blocks, one for each
of the eight possible waveforms defined by the user or system.
Apart from this memory configuration, this synthesizer
is particularly interesting because the oscillators may be
used to generate sounds according to five different synthesis
modes: fixed waveform, frequency modulation, VOSIM,
additive synthesis, and waveshaping2. This goes a long way
towards a "universal module" -that is, all modules of a
uniform type (with the resulting ease of conceptualization
and communication). While this is in direct contrast with
analogue synthesizers, a very wide repertoire of sounds is
possible (including all phonemes in Indo-European languages,
for example). We shall now present in greater detail the
actual design of this device.
2. Technical Details
In this section, we shall give the design details of the
digital synthesizer. Details will not, however, be taken to the
logic level. Rather, the purpose is to illustrate and discuss the
design approach to a level that enables the reader to evaluate
the appropriateness of this design as compared to the
alternatives. We shall begin by presenting an overview of the
general architecture. This is followed by a discussion of the
method of frequency control. Finally, a presentation of the
various acoustic models embodied in the design is made.
2.1 General Architecture
The general layout of the device is shown in Figure 1.
The synthesizer itself is made up of four main modules: the
controller, memory, oscillator, and digital-to-analogue
converter modules, respectively. Communication among the
Reprinted from Computer Music Journal, Vol. 2, No. 4 (1978).
46
