Beginner’s Guide: Everything you need to know about synthesis in music production

Synthesis and sound design are vital skills for any music producer. But there’s more to it than fiddling with knobs.

Beginner's Guide To Synthesis

Image: Jose Carlos Cerdeno Martinez

Synthesis has evolved exponentially in the almost 60 years since Robert Moog and Don Buchla unleashed their modular synthesizers on the world in 1963 and 1964, respectively. We’ve gone from analogue to digital technology, from hardware to software, and from subtractive to modelled synthesis – and often right back again.

Here, we work through this tangled mass of interwoven threads to tease out the things you need to know to become a master of synthesis .

Subtractive synthesis

The first form of synthesis was subtractive. The idea behind subtractive synthesis is to start with a harmonically rich sound source – a sound containing lots of harmonics, AKA partials or overtones – and then subtract some of that harmonic complexity. This recipe is universally popular. It’s intuitive and easy to get to grips with. It’s capable of creating a wide range of voices. And most importantly it sounds awesome.

A subtractive synth’s sound source is called an oscillator. In the analogue domain, this is often referred to as a voltage-controlled oscillator, or VCO. This is because the frequency of the oscillator is controlled by a variable voltage called a control voltage, or CV. For this reason, you’ll often see VCOs in software-based subtractive synths.

An oscillator creates repeated cyclic voltage fluctuations in an electrical signal; the pattern of fluctuations is known as a waveform. Waveforms are named after the shapes they create as seen on an oscilloscope. In sound synthesis, the most commonly used waveforms are triangle, square, pulse and sawtooth. Different waveforms result in sounds with different harmonic content, and therefore each has its own distinct timbre. Some oscillators can also produce sine waves but, because a sine wave is a pure tone containing no harmonics, they are of limited use in subtractive synthesis.

Fig.1 - Standard subtractive waveforms
Fig.1 – Standard subtractive waveforms

Subtractive synthesis is achieved with a synth’s filter(s), which removes some of the harmonics from the sound, thereby changing its timbre. There are many filter types. The most common is the low-pass filter, or LPF, so called because it allows frequencies below the filter’s cutoff frequency to pass through unaffected, while harmonics of a higher frequency are attenuated.

A filter’s cutoff frequency is not a hard limit that prevents any higher frequencies from passing through. Instead, the further above the cutoff a frequency lies, the more it will be attenuated, resulting in a slope-off of the frequencies above the cutoff. We can represent this using a filter curve graph – see Fig.2.

Fig.2 - Filter types
Fig.2 – Filter types

The other common filter types are high-pass (HPF), band-pass (BPF) and notch (AKA band-eliminate/band-reject – BEF/BRF). As with the LPF, the names of these other filters describe their function.

There’s also the comb filter, which is like a stack of notch filters, each with its own cutoff frequency. Conceptually, it’s like the teeth of a hair comb.

As well as reducing harmonics, most synth filters also feature a resonance control that accentuates the harmonics around the cutoff frequency (see Fig.3).

Fig.3 - Filter resonance
Fig.3 – Filter resonance

The final essential element of subtractive synths is the amplifier, known in the analogue world as a voltage-controlled amplifier, or VCA. As well as determining the volume level of a synth’s output, the VCA also controls the way the synth’s volume changes over time, from note on to note off, using an envelope. Envelopes are a form of modulation, a core concept in all synthesis that we’ll come back to shortly.

Learn more about subtractive synthesis here.

Additive synthesis

The opposite of subtractive synthesis is additive synthesis, which builds a complex sound from sine waves, which are harmonically pure waveforms. All sound can be disassembled into stacks of sine waves of different frequencies and amplitudes. So, in theory, additive synthesis can create any and every conceivable sound.

In practice, however, the sheer number of sine waves required to create even something as simple as a monophonic square wave makes additive synthesis quite complicated. Historically, assembling the sufficient processing power to create this plethora of sine wave oscillators has been challenging. However, with modern computers, this is no longer an issue. What remains an issue, though, is the complexity of managing and controlling a stack of sine waves to get the desired results.

If you want to own a classic additive synth, the Kawai K5 is probably the most realistic proposition. But these are rare, awful to program, and rarely give results that are worth the considerable effort involved. If you want to experience a workable, up-to-date additive synth, the VirSyn Cube and Cube 2 are good examples of how modern computing power can be used to harness at least some of the promise of additive synthesis.

Learn more about additive synthesis here.

FM synthesis

We’re all familiar with the vibrato effect, whereby the pitch of a note moves up and down. As long as the pitch fluctuations occur at a rate below the lowest frequency that humans can hear (about 20 cycles per second, or 20Hz), all we hear is vibrato.

However, if we increase the vibrato frequency into pitched territory, we will no longer hear vibrato. Instead, we will perceive changes in sound timbre. In other words, new harmonics are created. This is the essence of frequency modulation, or FM, synthesis.

FM is similar to additive synthesis in that harmonic complexity is built from harmonically simple sources. But FM synthesis doesn’t stack sine waves atop each other. Instead, it uses one sine wave to distort another. Because of this, FM can create harmonic complexities with far fewer sine waves than additive synthesis, and much more manageable as a result.

Fig.4 - FM operator

The basic building block of FM synths is the operator. Operators are a combination of oscillator and amplifier, with an envelope to control that amplifier (see Fig.4). If an operator is connected to another operator, it’s called a modulator; if connected to the synth’s output (so that it can be heard), it’s called a carrier.

The number of operators offered by FM synths varies but it’s usually six or four. While the operators in the earliest FM synths only produced sine waves, some later implementations have a variety of waveforms.

The way the operators in FM synths are connected is referred to as an algorithm, which has a huge bearing on the sound that will be created – Fig.5 shows a few examples.

Each block in the diagram represents an operator, and connections run from top to bottom. The lowest operators in each algorithm are the carriers; the remaining operators in an algorithm are the modulators that alter the frequency of the connected operator. Generally, at least one operator can be fed back on itself too.

Fig.5 - FM Algorithm Examples
Fig.5 – FM Algorithm Examples

FM synthesis has a reputation for being complicated. The old Yamaha DX synths, for example, whose only window into their myriad settings was a tiny two-line LCD screen, were very awkward to program. However, this is less of a problem with today’s software-based FM synths. The first step in the FM learning curve is steep but, once you’re over the hump, you’ll find FM much more comfortable.

Learn more about FM synthesis here.

Phase distortion synthesis

Yamaha acquired an exclusive license in 1973 to commercialise Stanford University’s FM synthesis technology so that no other manufacturer could create an exact copy. So Casio, who wanted a slice of the 1980s digital synth revolution, developed its own variation: phase distortion (PD).

Casio CZ-1000
Casio CZ-1000. Image: Reverb.com

Sound-wise, PD is not dissimilar to FM but it tends to have a warmer, fuzzier tone. PD is generally easier to work with than FM too, largely due to PD’s slightly reduced versatility. Sadly, PD is now all but obsolete. There are a few PD plug-in synths available online but, for a true PD experience, hunt down an original Casio CZ or VZ synth.

Learn more about phase distortion synthesis here.

Wavetable synthesis

Wavetable synthesis was the brainchild of synth-design legend Wolfgang Palm. The system is based on the subtractive-synthesis model but uses a digital wavetable oscillator in place of a VCO.

Wavetable oscillators can load data sets (wavetables), tables of data in which each entry is a digitised waveform that differs from those above and below. With careful planning and calculation, it’s possible to devise a series of waveforms that morph smoothly from one shape to another. Fig.6 illustrates this with a series that starts with a triangle wave and ends in a sawtooth.

Fig.6 - Wavetable Illustration
Fig.6 – Wavetable Illustration

Thanks to the sheer number of waveforms and the ability to modulate the table position as a wavetable synth’s oscillator plays, this form of synthesis offers phenomenal flexibility with grand scope for timbral variety and expression.

Palm’s original PPG Wave models remain highly prized and valuable; Waldorf, who acquired PPG’s intellectual property, have received praise for their line of wavetable instruments; and one of the most popular and widely-used plug-in synths, Native Instruments Massive, has wavetables at its heart.

Learn more about wavetable synthesis here.

Sample and synthesis

Sample-and-synthesis, or S&S, is a catch-all term that describes the sample-based instruments that came along in the mid-1980s. Though specific implementations vary, most S&S synths are based on subtractive concepts, with oscillators replaced by a sample-playback engine.

S&S synths are essentially samplers with preset, fixed samples, hence also being known as ROMplers. This means they can convincingly emulate real-world acoustic instruments without demanding significant effort from the user. It remains a popular form of synthesis today, particularly in instruments aimed at performing musicians who favour convenience, realism and quality of sound over flexibility and individuality.

Roland D-50
Roland D-50. Image: Thomann

Excellent S&S plug-ins include IK Multimedia’s Sampletank. Here, though, the line between S&S and true sampler is blurred, and many instruments that could be described as S&S synths are in fact just sample banks built on top of a full-blooded software sampler such as Native Instruments’ Kontakt or Steinberg’s HALion.

Learn more about sample-based synthesis here.

Granular synthesis

Granular synths use samples as the basis of their sound. But these synths focus on snippets of the source sample – referred to as grains – that are looped in order to create intriguing tones and timbres. Moving the granular sampling point within the source sample reveals entire worlds of. Most granular synths pass this near-infinitely variable sound source through a subtractive engine for further shaping and sculpting.

Granular synthesis is superb for producing pad sounds and ethereal textures, and exploring the range of tones you can squeeze from even the most mundane of source samples is always enjoyable. However, granular synthesis demands something of a trial-and-error approach and so isn’t best suited to creating specific sounds that you may have heard or imagined.

Physical modelling

The aim of physical modelling is to emulate acoustic instruments – not by trying to recreate the sound those instruments make but by mimicking the physics that makes those instruments sound as they do – the plucking of a piano string, a beater striking a bell, etc.

Once you have a model, you can create and edit sounds using real-world concepts, including the length and tension of a string, or the shape and size of a drum.

Yamaha VL-1
Yamaha VL-1. Image: Yamaha

After more than two decades of research into physical modelling, Stanford University once again buddied-up with Yamaha in an attempt to recreate the runaway success of its FM collaboration. The result was 1994’s Yamaha VL-1.

Although only a monophonic instrument, the VL-1 was capable of producing string, brass and reed instrument sounds whose realism was unparalleled for the time and remains impressive today. The VL-1 didn’t enjoy the success of the previous Yamaha/Stanford tie-up. One reason for that was the arrival of a different type of modelling synthesis…

Learn more about physical modelling synthesis here.

Analogue modelling

Analogue modelling does with circuitry what physical modelling does with musical instruments. The most common approach is to mimic the behaviour of analogue circuits, in some cases right down to the level of individual components. This allows the creation of exceptionally accurate digital recreations of classic analogue synths and entirely new digital synths that combine components in ways that would be impractical in the world of analogue.

Vintage synth lovers will argue that analogue-modelled synths don’t sound as authentic as the real deal. But this method of synthesis offers convenience and affordability, especially in software synths.

Physical and analogue modelling are fundamental components of many current flagship hardware synths, and the latter is used in pretty much all plug-in subtractive synths on the market.

Learn more about virtual analogue synthesis here.

Modulation matters

One of the most important concepts in synthesis is modulation, which is used to automatically modify synth parameters to create movement and flair in patches.

A modulation source creates the modulation signal; the modulation destination is the parameter that is being adjusted. Typically, each modulation mapping of source to destination also includes a way of setting the strength or magnitude of the modulation to control its impact on the destination parameter.

There are three basic types of modulation. Triggered modulation starts in response to an event – such as the pressing of a note – and then continues until the event is finished. Envelopes are the most common example of this: playing a note triggers the envelope’s attack stage, and the envelope will continue to move through its stages until you release the note, thereby triggering the envelope’s release stage (see Fig.5).

Fig.7 - Standard ADSR Envelope
Fig.7 – Standard ADSR Envelope

Continuous modulation runs all the time. The most common example is the Low Frequency Oscillator, or LFO. These are just like regular oscillators but run at much lower, sub-audible, frequencies, and will typically run at a fixed frequency. Most synths will have at least one LFO source, and many subtractive synths will have an oscillator, often featuring a sine wave, that can be switched into low-frequency mode.

Realtime modulation is generated by your performance while playing a synth and includes functions such as velocity, aftertouch, pitch bend and mod wheel. As such, we use this type of modulation to create expressiveness and to mimic the performance nuances of acoustic instruments.

Synthesis and sound design are remarkably deep topics. There’s a lot to learn. The best way to improve your skills is through practice. Whatever your ability, our ever-growing collection of guides and tutorials is there to fill the gaps and inspire new ideas.