synthesis
[To find out about modular synthesis in particular, click here, or to hear music
and emulations of real life sounds that were created using it, click here.]

There are many different types of synthesis, such as modular synthesis and granular synthesis, but the most common is subtractive synthesis and that is the main type discussed here. It is so called because it is a method of subtracting overtones from an oscillator signal using at least one filter. There is a fixed signal path in most synthesisers and this 'path' starts with the oscillator(s). Their output(s) are fed to some kind of basic mixer, possibly with modulation options, then sent to a filter, usually with an envelope assigned to it, and then on to an amplifier with its own envelope. LFO's and possibly other envelopes can then modulate various aspects of the sound to enhance it. Below are explanations of the individual components used in synthesis, with included audio examples in .mp3 format.

oscillators

Oscillators are the main sound generating components in synthesisers.

Oscillators generate a variety of continuously repeating, frequency-stable waveforms and as such, are the main sound generating components in synthesisers. They can operate over a wide frequency range that covers most of the human range of hearing. Oscillators that repeat faster than ten times a second (10 Hz) are called audio frequency oscillators, as they are audible. Oscillators that work at lower rates are called low frequency oscillators (LFO). Most synthesiser oscillators can produce several different shaped output signals.

top

oscillators  •  oscillator shapes

Oscillator shape refers to the shape of the oscillator's output signal when plotted on a graph of level against time. There are five main waveforms, or shapes, generated by audio frequency oscillators. They are sine, triangle, saw-tooth, square and pulse) (or rectangle). They each have a particular timbre and these raw, unfiltered sounds can be heard playing arpeggios over 4 octaves by clicking on the various listen buttons below.

The graphs below all show one cycle of a waveform starting at zero amplitude, rising to positive maximum, through zero to negative maximum and then back to zero, except for the pulse waveform graph, which is always at either zero or positive maximum level.

oscillator shapes - one cycle of each waveform

oscillator shapes  •  sine

The sine waveform produces a pure tone at the fundamental frequency - it has no overtones. This pure tone quality makes it excellent for bass drum sounds and as an element of bass sounds. Most sounds in nature are thought to be comprised from extremely complex combinations of sine waves.

oscillator shapes  •  triangle

The triangle waveform exhibits a slightly brighter tone than a sine wave and features some lower order overtones. It is also particularly good for low frequency sounds, where the overtones can often produce a slight gritty quality. This can give the crack of a bass drum a different flavour.

oscillator shapes  •  saw-tooth (saw)

The saw-tooth, or saw waveform gets its name from its shape and has a lively buzz to it. It is one of the brighter sounding waveforms, and it produces a great number of overtones. It is therefore regularly used for lead sounds and string or pad sounds. At very low pitches, the oscillator signal's jump from minimum to maximum (or vice versa) becomes very evident and can be used to good effect.

oscillator shapes  •  square

Of these common oscillator shapes, the square waveform produces the brightest sound, with the largest number of overtones. It sounds like a bright sine or triangle wave, having more body than a saw-tooth and therefore excels at creating both bass and lead sounds. A square wave is very similar to a pulse waveform with exactly a 50% duty cycle, except that it is negative for half of the cycle.

oscillator shapes  •  pulse

A pulse waveform can be thought of as having a variable rectangular shape and can have its duty cycle altered by a shape or pulse width control, usually between 50% (square wave) to 100% (virtually no output), with the sound progressively thinning or hollowing as the width is raised. The pulse width can be thought of as altering the length of time that the waveform sounds within each cycle. When this is modulated by an LFO, it is called pulse width modulation (PWM) and produces a fattening, chorus-like effect, similar to two slightly detuned oscillators. To hear the sound of a pulse waveform playing an arpeggio and changing from a 50% to a 93% duty cycle and back, click the listen button below.

[Note how the sound thins out as the duty cycle approaches 100%.]
top

oscillators  •  noise oscillators

When added to a signal at low to mid levels, noise can make the sound appear brighter.

Many synthesisers provide a noise oscillator as well as the regular waveform ones. White noise is a mixture of sound waves that evenly cover the entire frequency spectrum. Pink noise is also common and is like white noise except that it's adjusted to have equal energy per octave, resulting in a more natural frequency slope with fewer high frequencies. Some synthesisers also offer a colour control to select a brightness for the noise. When added to a signal at low to mid levels, noise can make the sound appear brighter. It is also commonly used in conjunction with filter sweeps for intro sounds that go 'whoosh'. Two basic examples of this can be heard by clicking the listen button.

oscillators  •  low frequency oscillators (lfo)

LFO's are used to modulate various qualities of the sound being produced.

If the rate at which a shape generating oscillator repeats is less than ten times a second (10 Hz), then it is considered to be a low frequency oscillator (LFO), as it is under the lower limit of human hearing. LFO's can therefore not be heard, but are used instead to modulate various qualities of the sound being produced. Some LFO's have a contour control that has the effect of biasing the shape it generates to create a wider range of modulation patterns. See the cutoff frequency section to hear an example of an LFO at work.

oscillators  •  oscillator sync

...changes to the slave oscillator's pitch will produce changes in the timbre of the sound.

Many synthesisers allow one or more of their oscillators to be synchronised with the output signal from another oscillator. Synchronisation forces the modulated (slave) oscillator to restart its waveform every time the modulating (master) oscillator reaches the start of its cycle. The effect of this is that the synced oscillator will inherit the pitch of the master oscillator(s) and form a complex waveform dependant on both or all of the oscillator pitch(es). The fundamental frequency will come from the master oscillator's pitch, while changes to the slave oscillator's pitch will produce changes in the timbre of the sound. Great effects can be achieved by sweeping the slave oscillator's pitch with an envelope or LFO. Click the listen button below to hear an example of this.

oscillators  •  detuned oscillators

Slightly differing, or detuning the pitches of two or more oscillators can thicken the produced sound.

Slightly differing, or detuning the pitches of two or more oscillators creates a chorus-like effect, which can thicken the produced sound. Increasing the difference between the oscillators' pitches will speed up the effect and make the sound appear fatter (up until a point when the result just sounds horribly out of tune). A side effect of this is that the high frequency content in the signal is reduced, becoming less focused, as the oscillators become more detuned. Modulating the fine pitch or detune controls with an LFO can add to the thickening effect. Click on the listen button to hear an example of two unfiltered square wave oscillators detuned by 6, then 12 and then by 24 cents.

top

filters

...filters attenuate parts of an input signal's frequency spectrum.

There are many different kinds of audio filters, but they all attenuate parts of an input signal's frequency spectrum, depending on the filter type, the cutoff frequency and the resonance settings. As well as the different types, filters can also have different slopes, causing varying degrees of attenuation. On top of all this, variations in the actual electronic components that make up filters will give each brand its own unique 'flavour'. True analogue filters tend to sound the best, although some modern computer and synthesiser emulations of them can sound very similar.

The filter banner below shows basic graphs of frequency, plotted from low to high (left to right), against level plotted from low to high (bottom to top). By clicking on the various listen buttons underneath the banner, you can hear examples of the various topics below.

Filter graphs
filter graphs - frequency plotted against level (dB)

filters  •  cutoff frequency

As just mentioned, filters attenuate parts of an input signal's frequency spectrum. The filter's cutoff frequency determines which of the input signal's frequencies are cut and which remain unaltered. Rather than cutting off all signals past the cutoff frequency, gentle slopes are employed, so that frequencies nearer the cutoff are attenuated less and those further away are attenuated more. The cutoff frequency is actually the frequency near the top of the slope at which signal content is 3 dB quieter than the unattenuated portion of the signal. Click the listen button to hear an example of a LPF with its cutoff frequency being raised by an envelope.

[Note how the output level increases as the filter opens and allows more frequencies to pass.]

filters  •  resonance

Virtually all filters used for synthesis include a feedback loop that introduces resonance, which has the effect of emphasising a band of frequencies around the cutoff frequency. High resonance levels also have the effect of reducing the level of all frequencies away from the cutoff. This generally makes the sound appear smaller, unless the cutoff is in the bass region, when the resonance boost can easily raise the overall level. At full resonance, some filters can self oscillate - that is, they produce their own tone, similar to a sine wave. Click the listen button to hear an example of a LPF cutoff being modulated by an envelope, as the resonance is raised to self oscillation.

[Note how the output level decreases as the resonance level increases.]

filters  •  filter types

There are many types of audio filters and they all attenuate certain frequencies, depending on the filter type, the cutoff frequency and the resonance settings. The main ones used for synthesis are low pass (LPF) which cut high frequencies, high pass (HPF) which cut low frequencies, band pass (BPF) which cut frequencies both higher and lower than the cutoff frequency and band reject (BRF) which cut those near the cutoff frequency. In the audio examples below, a loop can be heard first with a LPF, then a BPF, then a HPF and then finally, a BRF. All of the examples have a cutoff frequency that is modulated by an LFO to give an idea of how the different filter types sound at different frequencies.

[Note how the modulated cutoff frequency on the BRF example sounds similar to a phaser effect.]

filters  •  filter slopes

Each of the different filter types available can also have different gradient slopes. They are measured in steps of 6 dB/octave and a filter with this gradient is called a 1 pole filter, ie. 2 pole = 12 dB/octave, 3 pole = 18 dB/octave, etc. The steeper the filter slope, the greater the level of attenuation that occurs past the cutoff frequency, ie. using a 24 dB/octave filter can make a sound appear smaller than using a 12 dB/octave filter, as frequencies past the cutoff are attenuated by twice as much. Click the listen button to hear examples of different filter slopes. First a loop is heard with a 2 pole LPF, followed by a 4 pole LPF. The next two examples are of the same loop with a 2 pole HPF and finally, a 4 pole HPF.

[Note how the sound appears slightly smaller, yet 'tighter' and more focused in the two 4 pole examples.]
top

envelopes

Envelopes are one of the main sources of modulation, or sound shaping, available in synthesis.

Envelopes are one of the main sources of modulation, or sound shaping, available in synthesis. Offering a number of controls, they allow precise level adjustments to be made to various elements of the sound over time, ie. a filter envelope will change the cutoff frequency and an amplifer envelope will change the amplitude level. It is quite common for synthesisers to have envelopes 'hard wired' to the filter and amplifier and often provide additional envelopes to modulate other selectable parameters, such as filter resonance, or pitch.

They are triggered by a MIDI note on signal, ie. when a key is pressed on a MIDI keyboard, or a note is sent from a sequencer. The most common envelopes are known as ADSR's and have four stages, attack, decay, sustain and release, but it is also possible to get envelopes with more, or less stages. Multi-stage envelopes provide extra stages, such as second attack, decay and release stages and sometimes extra level controls, such as start and end levels.

The envelope banner below show the four stages of an ADSR envelope on basic graphs, with time plotted from low to high (left to right) against level plotted from low to high (from bottom to top) and an example of the extra stages on a typical multi-stage envelope. Click on the various listen buttons underneath the banner to hear examples of different envelope parameter settings.

Envelope stages
adsr envelope stages - attack, decay, sustain and release and a multi stage envelope (highlighting additional stages)

envelopes  •  attack

The attack parameter on an envelope determines the amount of time that its output level will take to rise from zero to maximum. Once the level is at maximum, the attack stage is complete. A long attack time on an amplitude envelope is needed to create a string or pad sound, while percussive sounds like drums usually require an instantaneous attack, ie. set to minimum. The .mp3 example below is of a repeated loop with a modulated attack time on the filter cutoff, starting at zero and ending at a setting of around half a second.

[Note how the filter cutoff sounds lower when the attack time is raised, because the following notes reset the envelope before it has enough time to reach the end of the attack stage.]

envelopes  •  decay

The decay stage begins once the attack stage has finished and the output level is at maximum. Its setting determines the amount of time that the output level will take to fall from maximum to the level that has been set on the sustain parameter. Generally, short decay times are needed for percussive sounds, while long decay times are better suited for slowly evolving sounds. The .mp3 example is of a loop with a decay time on both the filter and amplifier envelopes, changing from 100 ms to around 1.4 seconds.

[Note how the cutoff remains higher when the decay time is raised, because the next note resets the envelope before it has enough time to drop to the level set on the sustain parameter.]

envelopes  •  sustain

Sustain determines the level that an envelope will fall to once the decay stage is complete, if and only if, a note is played longer than the sum of the attack and decay times. The length of a sound with an amplifier envelope sustain level of zero will be the sum of the attack and decay times. A sound with an amplifier envelope sustain level at more than zero will be continuous, ie. percussive sounds benefit from having a sustain level of zero. The .mp3 example below is of a loop with the sustain parameter on both the filter and amplifier envelopes, changing from a level just above the minimum up to the maximum level.

[Note how the decay stage 'vanishes' as the sustain level reaches maximum.]

envelopes  •  release

The release parameter determines the amount of time that the output level of an envelope will take to fall from the level that has been set on the sustain parameter to zero after a note has ended. String and pad sounds generally use long release times, while percussive sounds need short release times. Using an amp envelope with a medium attack, no sustain, zero decay and zero release times can create a pleasing backward sounding effect. The .mp3 example is of a repeated loop with a modulated release time on the filter and amplifier envelopes, starting at 50 ms and ending at a setting of 1.6 seconds.

[Note how it is the end of the sound getting longer and the decay stage remains audible.]

envelopes  •  envelope amount

The amount, or envelope amount, control on an envelope determines the amount of the envelope signal that is fed to its output. When the amount parameter is set to maximum, the envelope will operate over the full range of levels of whatever parameter it is modulating. With any setting less than maximum, the envelope's output signal is reduced and any level settings on the other parameters, such as sustain, are scaled down so that the correct envelope shape is maintained.

Envelopes 'hard-wired' to amplifiers generally have no envelope amount control, as reducing this would lower the output level of the amplifer...

An amount value of 50%, or half, would restrict the modulation range of the envelope to half of its maximum range. For example, if an envelope output is modulating a filter's cutoff frequency and its amount is set to 50%, it will raise the cutoff by half of its range when it is at its highest point (once the attack stage is complete). Envelopes that are 'hard-wired' to amplifiers generally have no envelope amount control, as reducing this would lower the output level of the envelope and therefore also the amplifer, ie. the output volume of the sound being synthesised.

To find the optimal envelope amount setting, set it first to zero - this allows you to hear the sound that will be heard when the envelope is at its minimum, ie. start and end of envelope. Adjust the modulated parameter (cutoff frequency, pitch, etc.), until you are happy with the sound of the lowest level that the modulated parameter will reach. Adjusting the envelope amount parameter will determine the highest level that the modulated parameter will reach. Generally, sounds with shorter decay times benifit from smaller envelope amount settings. The .mp3 example below is of a loop with the envelope amount parameter on the filter envelope, changing from the minimum up to the maximum level.

top

modulation

Modulation refers to the changing in level of a parameter over time.

Modulation can bring static sounds to life. It basically refers to the changing in level of a parameter over time. Whether it's a sine wave from an LFO modulating the fine tune or pulse width control on an oscillator, or an envelope creating a long filter sweep, you'll find sounds much more interesting when they have movement in them. All decent synthesisers offer at least two envelopes and two LFO's for this purpose. Modulation source refers to the LFO, envelope, modulation wheel on a keyboard, or whatever is generating the modulating signal and modulation destination refers to the element of sound being modulated, ie. an amplifier, a filter, a pulse width control, etc.

modulation  •  frequency modulation (fm)

Frequency modulation (FM) is actually a different kind of synthesis, but many subtractive synthesisers offer a basic version of it. It is a type of synthesis where the frequency of a carrier wave is altered by a modulating signal from another oscillator. At low levels, it can have a nice thickening effect and at high levels, it creates highly complex and bright waveforms which are very good for creating inharmonic and metallic sounds, ie. drums, bells, etc. For an FM sound to be in tune, both oscillators need to have whole octave pitch differences. FM can also create interesting effects... the first audio example is of an unfiltered saw wave frequency modulating another, while an envelope is raising the modulating oscillator's frequency and the others are of a variety of unfiltered FM sounds.

modulation  •  amplitude modulation (am)

Amplitude modulation (AM) is similar to frequency modulation except that the amplitude (and not the frequency) of a carrier wave is altered by a modulating signal from another oscillator. It produces an audible sideband on either side of the carrier frequency. When the modulating oscillator's frequency is higher than the carrier frequency, the fundamental frequency of the carrier wave remains audible, but when it is low, the modulating signal is heard more as amplitude level changes of the kind an LFO would make on an amplifier. The first unfiltered audio example is of a low carrier frequency with a falling modulating frequency, the second is of a higher carrier frequency with a lower, falling modulating frequency, the third is of a high carrier frequency with a modulating frequency rising from low and the last example is of a low, rising carrier frequency with a low static modulating frequency.

modulation  •  ring modulation

Ring modulation, more commonly known simply as ring mod, is related to amplitude modulation. It is a process where two waveforms are combined by multiplying and the sum and difference between the two is output. It can create inharmonic waveforms that are good for synthesising metallic and bell-like sounds. Click on the listen button to hear a variety of typical sounds created using ring mod, with the final sound including a frequency sweep to allow you to hear a range of timbres available.

top

Many synthesisers also have effects sections... to find out about the different kinds of effects with audio examples, click here.

 

 

Copyright © 2015 nitetime productions. All rights reserved.