of Microprocessors

SIGNAL ROUTING 287
Increasing Flexibility
A synthesizer based on such specific fixed function modules is lirtle
more than a computer-controlled organ with a wider than normal selection of
stops. More flexibility can be had by making the previously fixed elements
variable under computer control. For example, the fixed filter could be made
variable as could the waveshaper. Some or all of the envelope parameters
could also be made variable. With these changes, a single module type could
cover a range of instrument types with appropriate control from the computer.
An even more generalized module could be imagined. A selector might
be added to the VCO so that any of the standard VCO waveshapes could be
selected. Ultimately, a programmable waveform generator (in which a small
memory loaded by the computer contains the waveform) driven by the VCO
would be used. The filter could be expanded to a full VCF with selection of
the filtering function as well as the cutoff frequency and Q factor. The
envelope generator could also have a programmable envelope shape like the
waveform generator.
Even with voice modules of the type shown in Fig. 8-6 in which every
parameter can be controlled, there are still serious t1exibility limitations. For
example, a sound may require two independently tunable formants, which
requires two bandpass filters. FM synthesis is not possible because only one
VCO is present. In fact, the majority of special effects is just not available
with the example module.
The situation can be improved markedly, however, by expanding the
module somewhat and allowing perhaps a half-dozen different configurations
through a simple switching arrangement. Commercial voice-modular synthesizers
typically include two independent VCOs, two variable filters,
several VCAs, and several envelope generators, the latter usually implemented
in software. Available interconnection patterns range from two
parallel voices, to various intertwined combinations of the elements, even to
FM connections where one VCO modulates the other or one of the filters. In
particular, the two VCOs have been found to be very flexible, often
eliminating the need for a generalized waveshaper (the second oscillator can
be tuned to provide a prominent harmonic, for example). A moderately
effective chorus effect can be obtained by having the VCOs almost track each
other. A specific example of a voice modular synthesizer using these
techniques can be found in Chapter 19.
While these measures are typically more than adequate for a performance-oriented
synthesizer, t1exibility for special cases is still limited. Thus, a
few independent function modules and processor modules (like frequency
shifters, chorus generators, and reverberation devices) are needed to supplement
the voice modules if a flexible, comprehensive system is the goal.