Chapter Four: Synthesis
4. Important analog synthesis concepts
As the name implies, sounds are created synthetically via electronic circuits, as opposed to originating from sampled, or real-world recordings. Voltage is used to control virtually all functions of the instrument, including pitch, timbre, amplitude over time, vibrato, and tremolo. Common controlled modules are referred to as:
VCO: voltage-controlled oscillator
VCF: voltage-controlled filter
VCA: voltage-controlled amplifier
The 1964 Moog Modular synthesizer set the gold standard for voltage-controlled synthesis in a single instrument design, and most software synthesizers (such as Native Instruments Absynth), and even synthesis programming languages these days still follow the basic architecture of that instrument. Genius designer/inventor Robert Moog's instrument was highly configurable via patching, or connecting modules via patch cords. It was not unusual to have dozens of patch cords routing audio and control signals to create a single sound. Subsequent lower-tier Moog models, such as the MiniMoog hardwired some of the more common connections and were not as configurable.
Pictured left: Moog Model 55 (1974), a model once owned by Indiana University (a sad story...it was serial #5, built in his garage...)
Pictured right: Moog 921 Voltage Controlled Oscillator module (click for larger image)
While many of the control connections are often not as flexible in some software programs, you can still create your own instrument designs (still called patches!) with graphic synthesis programs such as MAX/MSP, which is infinitely configurable via drawn patch cords between graphic objects which themselves are the equivalent of hardware modules (MAX actually comes with a set of analog-style synthesis modules called BEAP—Berklee Electro-Acoustic Pedagogy), and you can follow along with this lesson if you have access to MAX.
Most voltage-controlled synthesizers are modelled around subtractive synthesis techniques, where the sound path begins with a waveform (or sample these days) rich in partials. These frequencies are then partially removed (subtracted), or otherwise shaped, by a filter. The antithesis of subtractive synthesis is additive synthesis, where sound is produced by summing numerous sine waves together to create composite timbres, and with no filtering.
Functions of Control Voltage (c.v.)
Without detailed electrical engineering knowledge of the nature of voltage, simply viewing it as a force applied to a task, such as changing the frequency of an oscillator, or making a sound louder or softer works well…the more voltage, the greater the effect. While the sound output we ultimately hear from a synthesizer is also output as audio-rate fluctuations in voltage, control voltage serves a different function, which is to modify the parameters of various synthesis modules, sometimes at audio-rate, sometimes at sub-audio rate, and sometimes as a fixed single value.
C.v. can be applied to affect many things. In the simplest patch (what we refer to as the basic patch on the following pages) it controls the pitch of an oscillator via a keyboard, it controls the cutoff frequency of a filter, it controls the amplitude of a sound over time (including start and end). It can also control pulse width, filter resonance, and much more. Control voltages can be summed with other c.v.’s. For example, vibrato is often created by summing control voltage, patched to a signal oscillator, from a keyboard providing the center pitch c.v. with a low frequency oscillator creating the pitch fluctuation around the center pitch. Traditionally, the range of voltage used in the classic voltage-controlled synthesizer is 0-5 volts (not nearly enough to electrocute, or I wouldn't be around to write this sentence).
|Some modules have several types of control inputs. For VCO's and VCF's, it was common to have three c.v. input types-1V/oct, linear and exponential. For both of those modules, 1V/oct would change the oscillator or filter cutoff frequency by an octave for each volt's change in c.v. and was not adjustable. The synth keyboards were calibrated to output a 1V change per octave to match this input. The VCO would therefore respond with equal tempered note spacing when 1V/oct was used. Linear moved the modules an equal number of c.p.s. up or down around a center frequency (crucial to "Chowning-style" audio-rate FM we will cover later) , while exponential moved these parameters an equal musical interval up and down. Unattenuated exponential was equivalent to 1V/oct.|
Signal Paths vs. Control Paths
When studying synthesis, it is important to keep in mind which connections are routing the audio signal, i.e. the signal we will ultimately hear at the end of the line, vs. control connections, which we will not hear as audio. Control connections affect what we hear, for example the pitch, but we don’t hear their signal directly as audio output. Some synthesizer modules can be used as a source for both audio signal AND control signal depending on how they are patched. The voltage-controlled oscillator, for example can provide either audio as part of the signal path, or be a source of control for tremolo or vibrato if it is patched to control inputs rather than the signal path.
Types and Sources of Control Voltage
AC control voltage—cyclic or noise-based control voltage from oscillator, frequently an LFO (low frequency oscillator), or noise generator
DC control voltage—single value or non-cyclic single stream control voltage from sources like keyboard, envelope generator and gate (usually one long pulse wave for duration of note generated by keyboard key on and key off)
Modulation is created by applying an AC control voltage, either from an oscillator or noise generator, to a module’s function, such as frequency, filter cutoff frequency, or amplitude. In addition, other voltage-controllable functions such as pulse width can be modulated (very cool effect), filter Q can be modulated, as so on. Three parameters are key to modulation: rate, depth and wave shape. Regarding wave shape, if a sine wave was applied to the frequency control input of an oscillator, a smooth up and down vibrato would ensue. However, if a square wave was applied, a trill would ensue.
Audio-rate vs. sub-audio rate modulation:
Audio rate modulation occurs when the control oscillator (or noise source) is set above ~20 Hz. This can often create additional frequencies, some called sidebands.
Sub-audio rate modulation, below ~20 Hz, is often accomplished with a low-frequency oscillator or LFO.
Important sub-audio modulation types are:
Tremolo: amplitude modulation
Vibrato: frequency modulation
Timbre modulation: filter cut-off frequency modulation (no good term for this, but the filter’s cutoff frequency is modulated), shifts the cut-off frequency higher and lower-more noticeable with higher resonance (Q) values
Pulse width modulation: by varying the pulse width of an oscillator’s pulse waveform, changes the number and strength of the waveform’s partials (very cool effect, try it)
Double Modulation—when one modulating oscillator’s frequency is modulated by a second control oscillator.
The following two terms are essential to understanding 99% of a voltage-controlled synthesizer’s settings and are common to most signal-path modules such as oscillators, filters and amplifiers:
Offset: sets the initial setting of a module, such as the initial frequency of a VCO, the initial cutoff frequency of a VCF or the initial amplitude of a
VCA. This is where the module’s parameter begins before control voltage is applied.* For example, if you play middle C on a synthesizer keyboard and that sends out three volts to the frequency control input off an oscillator, you can offset or tune the oscillator to whatever pitch you want, not necessarily middle C. If you then play a C an octave higher, and the oscillator receives one additional volt, if the control input is set to respond 1 volt per octave (1V/oct), the oscillator will output an octave above of whatever you set with the offset.
An interesting example of an offset was Irving Berlin’s transposing piano, which had a shift lever to move the mechanism to whatever key he wanted to play in, as the famous, but self-trained composer could only play in the key of F#.
*The Absynth soft-synth we use takes a different approach, whereby the offset is actually calibrated to be the maximum value of a parameter at maximum c.v. rather than the initial value at 0V c.v.
Attenuator: cuts down on the effect of applied control voltage, For example, a filter won’t sweep as wide a frequency range when an envelope generator is applied to it. On current soft-synths, often called “depth.” In the above offset example, while playing an octave higher on a keyboard may send an additional 1 volt, if it is attenuated, the module will only receive a portion of that voltage change, and therefore only move a fraction of the octave. If the attenuator reduced the 1-volt change to .5 volts, then the oscillator would shift a triton rather than an octave. The result would be a microtonal scale rather an equal tempered one. If the control voltage is completely attenuated, it won’t change frequency at all. In classic synths, many control inputs (except the fixed ones labeled 1V/oct) had attenuator knobs directly above or below the c.v. jacks, as depicted in the c.v. section above.
An envelope generator (EG) in the basic patch (also called an ADSR) creates a changing stream of DC control voltage across the duration of a note. In the basic patch, two EG’s are used. The first is applied to the VCA. It controls the onset of the note and its initial rise time to a maximum amplitude, decay time to a sustain level where it stays until a note is released, and then the final decay back to silence.
A second EG is frequently applied to a filter’s cutoff frequency and shapes the timbre over the course of a note in a similar fashion. Using two envelope generators allows the timbre to change independently of the amplitude.
|A classic envelope generator had 4 stages—pay close attention to the words ‘time’ and ‘level’ below. Newer software graphic envelopes have the capability for both time AND level for each stage.|
Attack time: the time it takes to rise from 0V to maximum c.v. level (usually +5 volts). If the attack time is set to 0, the note begins at maximum amplitude and goes immediately to the decay phase.
Decay time: after reaching the maximum attack LEVEL, the time it takes to decay to a sustain LEVEL
Sustain level: the amplitude remains constant until the envelope is ungated, at which point it moves to the release phase
Release time: The time is take to go from the sustain level back to 0V. If the release time is set to 0, the note ends abruptly.
|Modern EG’s, particularly graphic EG’s in software applications address some of the shortcomings of the classic EG. For example, as classic EG must rise to a maximum voltage no matter what, whereas modern EG’s allow partial rise, or even descent for the attack. A classic EG had only four stages with limited shapes, whereas current EG’s can have numerous stages and many shapes, including slopes of lines. A classic EG always had a release to 0V, whereas a modern EG can actually have a rise on release, as picture on the left.|
Some classic envelope shapes below:
|Instant attack, moderate decay to sustain, then slow release|
|Slow attack, instant decay to sustain, then slow release|
|Instant attack, fast decay to 0V sustain level—makes percussive|
|Slow attack, instant decay to 0V sustain|
In addition to the basic patch connection of EG's to VCO and VCF, envelopes can be effectively used to control pitch over time (often summed with keyboard voltage). Much to the chagrin of this trombonist, a quick down and up divit is often added to brass-like patches, or a fast downward EG gliss to below 20 Hz can make a very interesting percussive sound. EG's can control the depth of modulation over time when used in conjuction with an amplifier (see the FM patch diagram on this page).
How does an envelope generator know when to begin and end its stream of voltage? It traditionally uses a gate. When a key on the keyboard is depressed, not only does the keyboard send control voltage to the oscillator to specify the pitch, but it sends a second type of signal called a gate, to all EG’s participating in the note event. When the key is depressed (called “note on” the pulse begins its positive 5v phase and stays there until the key is released (called “note off”) where the pulse returns to 0v value. The keyboard gate signal is patched to both envelope generators, and they each move through their attack and decay times to rest at the sustain level until the envelop is ungated (“key off”), at which point they begin their decay phase. Sometimes an envelope may not make it to the sustain level before being ungated, at which point it jumps to the release phase from wherever it was, and that may cause a click if the discontinuity is too great.
A variation for repeated notes would be to apply an LFO to either gate or trigger the envelope. Because a trigger is a single fast pulse without extended duration, the ADSR usually skips the sustain stage when triggered.
Sample and Hold (S/H): A module that samples the instantaneous voltage of an input signal at regular intervals and holds it at that level until the next sample. It works very much like sampling in digital audio. An S/H module has two inputs—one for the signal to be sampled, the other accepting a pulse wave to time the taking of samples. The S/H module outputs can control things like a VCO frequency or a VCF cutoff frequency. If a sine, triangle or sawtooth wave is sampled at certain rates, scales are produced. If noise is sampled, as was the case with every 60’s-70’s movie depicting a computer in operation, a random string of voltages is output, making for a random sequence of notes if applied to a VCO.
See Page 2 for the Basic Patch
1 | 2