Free technical help for digital creators - music, video, art, photography, graphic & web design

Synthesis types by Matt Ottewill

synth modules
Before we can talk about synthesis types we need to define a few terms and concepts.
(NOTE: If you have not already done so, you may wish to read our articles on Sound theory first).

Terms & concepts

What is synthesis?

Sound synthesis is the process of using electronics to create an electrical pressure soundwave from scratch and then controlling and modifying it.

Oscillators

An Oscillator is an electronic sound source. It is the device which creates the electrical pressure soundwave in a synthesiser. Oscillators can be analogue or digital.

CV, gates & MIDI

Before digital control, analogue synthesisers used an interconnecting arrangement of controlled voltage/gate signals to trigger their various components. For example, a key/note played on a keyboard would send a control voltage to an oscillator to tell it what pitch to produce, and another control voltage to an envelope which in turn would "instruct" an amplifier "shape" the volume envelope of the sound as it emerged from the oscillator.

Since the early 1980s, MIDI has been the primary way to control and play a synthesiser from an external controller, and internal digital interconnection has replaced the internal CV/gate control signals.

Patches

In the early days of synthesis when only modular systems were available, sounds were created by connecting modules with patch cords and adjusting settings on each module. The settings and patch cord connections for a sound came to be referred to as patches. Recalling a patch was a laborious process, until performance synthesisers began to integrate micro-processor control and ROM memory was developed to store a patch configuration digitally. There is still no way to store a patch created with a modular analogue system.

Types / methods of synthesis

Many different types of synthesis have been invented since the first successful (and affordable) commercial analogue synthesisers in the late 1960"s. The following are perhaps the best known ...

  • Subtractive (also know as Analogue)
  • Additive
  • FM (Frequency Modulation)
  • S & S (Sample and Synthesis)
  • Wavetable (wave sequencing & vector synthesis)
  • Phase Distortion
  • Physical Modeling (also known as Soft Synthesis, or Virtual Synthesis)

A Doepfer modular analogue synthesiser.

Korg prophecy

Synthesis "physical" forms

Most synthesisers are realised in one of the following forms ...

  1. Analogue hardware modular - such systems combine individual hardware components each playing a specific role (oscillators, LFO's, filters, envelopes, amplifiers etc). Components can be swapped and connected in a variety of ways using patch cords. Control signals are passed between them with a secondary interconnection system of CV/gate signals. Modular systems, such as Moog or Euro-rack, are the obvious examples. Patches cannot be saved and recalled.
  2. Integrated analogue performance synthesiser - essentially a pre-packaged and connected modular synth in a single case with a keyboard. Signals are routed with switches, and if a memory/patch feature is included, patches can be saved and recalled.
  3. Analogue sound module - an analogue performance synthesiser without a keyboard, which is usually controlled and played via MIDI.
  4. Integrated digital performance synthesiser - essentially a pre-packaged and connected software synth, with controller knobs and switches, in a case with a keyboard. Sounds are generated, edited, save and recalled digitally.
  5. Digital sound module - a digital performance synthesiser without a keyboard, which is usually controlled and played via MIDI.
  6. Computer hosted software plug-ins - no hardware, just software emulations running on DSP or the CPU, usually played/controlled from an external MIDI or USB connected keyboard/controller. These can "emulate" any synthesis type.

Subtractive synthesis

Subtractive synthesis was the original synthesis methodology. Because it used analogue rather than digital components it is often called analogue synthesis.

The classic integrated subtractive analogue performance synth, the Moog Minimoog.

Korg prophecy

First developed to great commercial succes by Bob Moog in the 1960s, it involves using oscillators to create electrical pressure soundwaves which are then processed, or "modified" to alter their pitch, frequency content and amplitude over time. In essence, elements of the original waveform are "wobbled" and "subtracted".

There are now numerous software emulations of the designs produced in the 60's and 70's, offering improved reliability and tuning but sometimes lacking the character and warmth of the old electronics.

Click here for a detailed explanation of Subtractive Synthesis.

Additive synthesis

Additive synthesis emulates the way in which sound is created in nature. An additive synthesiser has multiple oscillators. Each oscillator creates a simple sine wave. The user can set the frequency and amplitude of each sine wave. By combining these sine waves together a complex waveform is created. This process emulates the process by which natural sound is created with multiple harmonics.

In principal, these synthesisers are powerful, but in practice it has been very difficult to design a usable and practical control interface.

Commercially available additive synthesisers have tended to be complex and difficult to operate and have therefore had limited success. Examples include the Kawai K5 (1987).

additive synthesis
Kawai K5

FM (Frequency Modulation)

FM was the first commercially successful, and affordable, digital synthesis type. It is similar to Additive synthesis in that it uses 6 sine wave creating oscillators, each of which can have their frequency, amplitude and envelope (volume over time) set by the user. In FM, oscillators are called Operators.

An example algorithm showing oscillator connections, and the Yamaha DX7 (1983), the first and most successful FM synthesiser

FM synthesis
DX7

It differs from additive in that rather than combining the sine waves together, the output of one operator is sent to modulate, or "wobble", the next. The second then modulates a third and so on until a complex waveform is produced by the final oscillator in the chain, the so-called Output Operator.

The oscillators can be arranged in various "patterns" called algorithms, each with a different series of connections. Operators can feedback on themselves too. Cheaper 4 operator FM synths were sold but these lack the timbral richness and complexity of 6 operator ones.

FM lives on in many physical modeling synths today, and perhaps the most well used one is Native Instruments FM7/8 plug-in which is essentially a DX7 in software with additional features.

S & S (Sample & synthesis)

In the late 1980"s there was a demand for an affordable synthesise methodology which could produce a range of analogue, digital and natural sounds. This led to the development of S & S synthesis by Roland in Japan.

Rather than using waveform creating oscillators, an S & S synthesiser uses pre-recorded samples stored in ROM (Read Only Memory) as its sound sources. Due to the expense of ROM memory chips, only 8 to 16Mb capacity chips were fitted in early devices. It was therefore not possible to use full length high quality samples if a large variety sounds needed to be available.

Examples of S & S synthesis include the Roland D50 (1987) shown here, and almost all multi-timbral GM synthesisers (such as the Roland XP10 and those integrated into affordable PC Sound-cards).

s & s synthesis

Roland D50

To overcome this problem just the attack and a portion of the sustain of the sound is stored. The sustain section is looped to create the illusion of a longer sample. This works well because most of the identifiable character of a sound is derived from its attack. However, the more slowly evolving timbre of the sustain and decay parts of some sounds (such as pianos) cannot be reproduced by looping a short sample, so the sustain sample is processed through a conventional subtractive architecture of filters and amplifiers which create the illusion of a natural evolving decay.

Despite the fact that recent devices have greater storage capacity and therefore can utilise longer and better quality samples, this synthesis type is perhaps best suited to creating new sounds by combining samples of real instruments with samples of subtractive or other electronic synthesis types.

At the cheaper end of the market S&S is suitable for PC soundcard sound modules, but professionals have also found devices such as the Roland JD-800 and JV1080 to be unique and powerful.

Wavetable, wave sequencing & vector synthesis

This form of synthesis is ideal for slow pad-like and rhythmical evolving sounds. The essential concept involves sequentially cycling through (wave sequencing) or morphing between (vector synthesis) a number of waveforms over time. Complex evolving and sounds and rhythms can be created in this way. The waveforms are usually stored digitally in a so called "wave table" and may be samples or digital "models".

The Korg Wavestation (shown here) featured both vector and wave sequencing.

Korg Wavestation

Popular wavetable synths include the Sequential Circuits Prophet VS (1986) and the Korg Wavestation (early 1990s). Both these synthesisers employed a joystick controller which controlled the vector mix of waves over time. With the joystick in the default central position, an equal mix of all waves is heard but as it is moved away from this position the mix changes. This mix can be automated over time with a mix envelope and each wave can be modulated differently to change, for example, its pitch.

Phase distortion (PD)

Phase distortion is a form of synthesis, which uses an algorithm to create a sine wave, and then uses a second algorithm to distort the "shape" of the sine wave to create a totally new waveform. A simple example of how distortion alters a waveform would be inputting an oscillator generating a sine wave to a mixer input channel and then overloading the channel by increasing the gain until it distorts. The resultant waveform will be similar to a sine wave, but will have added harmonics. As a result of being clipped, the wave will sound more like a square wave. If you use an algorithm to achieve this, this would be phase distortion.

Examples of phase distortion synths include the Casio CZ101.

Korg prophecy

Two cycles of a sine wave showing phase distortion

Phase distortion

Phase distortion was used extensively on the Casio CZ series of synthesisers, notably the CZ101. It is best used when added to other forms of synthesis, such as Frequency Modulation, where the ability to create waveforms which are variations of sine waves is particularly useful for creating new tones.

Physical modeling / Soft Synthesis / Virtual Synthesis / plug-ins

This relatively new form of synthesis utilises fast, powerful digital signal processors (DSPs), or the CPU of a computer, and high quality digital to analogue converters (D.A.Cs). The synthesiser will take the form of either a program, which may run on a computer (normally called a soft synth or plug-in ), or a dedicated hardware device in the form of a rack module, or keyboard synth.

Examples of early physical modeling hardware synthesisers include: Yamaha VL1 (1995) and, shown here, the Korg Prophecy (1996). The Prophecy can model subtractive and S&S synthesis, and acoustic brass, woodwind, keyboard and string instruments.

Korg prophecy

Native Instruments Massive - a software (plug-in) synthesiser

Korg prophecy

As a note event is triggered, the program uses DSP to mathematically calculate in real time the stream of digital data required to create the sound it will output through the D.A.Cs

Consider a sampler which can convert an incoming analogue electrical pressure soundwave into digital (binary) data, which it then stores. This sound can then be replayed with a connected MIDI keyboard. The sound can respond to performance changes in velocity and pitch bend thereby altering volume, brightness (with the use of a low pass filter) and pitch, but otherwise remains unchanged.

A physical modeling synthesiser is actually generating the binary data (sound) in real time, tailored and responsive to the nuances of the performance arriving from a controller (usually a MIDI keyboard). Its ability to sound organic and "realistic" is entirely dependent on the skill of the programmers. Additional hardware controllers (such as breath controllers, log wheels, pressure pads etc) can also effect the way the sound is produced.

In addition, PM synthesis can model elements of instruments and combine them in ways unheard of in nature: for example, a bowed trumpet or a snare drum that alters pitch or size in real time.

Physical modeling can also be used to emulate any other form of synthesis. The range of sounds it can create is determined not by the hardware (providing the DSP is powerful enough), but by the available software models. Thus we have many emulations of subtractive (Logics ES1 soft synth, Propellerheads Reason), FM (Native Instruments FM7 soft synth), additive and S&S synthesis (Spectrasonics Atmosphere.) etc.

Many plug-ins today are developed using dynamic convolution, a process whereby impulse responses (essentially short recordings of a device or instruments output) are analysed and the results used to create real-time emulations (programs).

View in Landscape mode or Swipe to scroll
Types of physical modeling synth
  Hardware Hardware Software Software
Type Module Keyboard Stand-alone Plug-in
Example Nord rack Roland V-synth NI Reaktor NI Reaktor