Electronic music - approaches to making sounds

Smaller components

Modulation for audio purposes

Modulation alters a signal waveform using a carrier waveform.

Most modulation is not enough to make very interesting sounds, but the theory is useful to know especially when building or understanding lower level things.

https://en.wikibooks.org/wiki/Sound_Synthesis_Theory/Modulation_Synthesis

AM

The simplest form is amplitude modulation, where the amplification of the original signal is varied according to the level of the carrier wave.

Effect-wise, tremolo is amplitude modulation with a carrier typically on the order of a few hundred ms. #Ring modulation can be seen as bipolar AM, but more on that below.

FM

Another form is frequency modulation, where a higher level in the carrier wave means an increase in the frequency of the output.

This is interesting enough that it became a form of sound synthesis in itself.

A fixed form would just vary the frequency you generate, but there are fancier variants that get more involved.

Effect-wise, vibrato is frequency modulation (with a slow carrier(verify)).

PM

PM is closely related to FM, and can be used to implement FM for most signals.

It can be explained from a few perspectives, e.g. depending on whether its a radio thing or a sample based thing.

Let's go with the sample, because it's intuitively closer to sound.

Consider a PCM sample that contains a single sine wave. Typical production would be to have a cursor read off samples at a regular speed.

PM basically adds a small, varying offset to the cursor. In this case, it corresponds directly to the phase of the sine wave we're producing.

The resulting modulation is a continuously varying pitch shift.

Intuitively it's a doppler effect, as if we're shaking the speaker really hard (for sine anyway).

In synths, the rate (how fast this offset changes) is typically fairly low, as is the depth (amplitude, i.e. how much the modulator changes the offset).

And a lot of early (and current) FM was implemented this way because not only is it cheaper to implement (if your setup is digital lookups, then it reduces the modulation problem to basically just an offset), it is in some ways more precise: true FM is more likely to drift and accumulate other small errors, more so when operators are at audio frequency(verify), which is more likely with more-operator FM.

The main difference is that PM changes frequency only while the slope changes, whereas FM responds to the actual sample value.

For smooth waveforms, "shape" and "change in slope" are closely related. Yet anything with immediate changes will work differently:

• squarewave / PWM modulator

in true FM it would alternate between a higher and lower frequency

in PM it will amount to a tick twice per waveform, and the same frequency

• triangle wave

in true FM this will sound like a siren, with slightly jerky changes

in PM it is two tones (basically what you were thinking for squarewave FM)

• sawtooth wave

in PM: like triangle in that it ticks, but because it's only a positive or only a negative slope, it adds a constant to the pitch (raising or lowering)

will also not work, because of essentially the same reason.

(You can get almost the same effect as those would have in true FM with some clevermness)

This is roughly why many Yamaha OP* chips had only sine wave as a modulator, and even the fancier ones

The main difference between actual FM and PM to do FM is the response to immediate steps(/pulses) in the modulation waveform. This is why waveforms in FM, even if they aren't just sine, are typically controlled.

https://moinsound.wordpress.com/2011/03/04/frequency-modulation-or-phase-modulation-synthesizer-technologies/

Ring modulation

What is it?

tl;dr: RM is bipolar AM

Consider tremolo, which varies how loud something is. It is amplitude modulation (AM) of a potentially complex signal, with a slow sine wave going between 1 (as-it-was) to something quieter (potentially to 0, silent, but this is often too jarring).

Tremolo also means doing this slowly, because the faster it goes, the less noticeable it is.

Ring modulation starts off similar, but does two things differently:

One, it effectively multiplies with -1 .. 1 instead of just 0..1, in other words, it inverts the input half the time.

When it flips between inverted and not, it effectively introduces phase differences between those two parts, which is roughly the reason that, unlike tremolo, it introduces new harmonics.

The other is that it keeps doing interesting things with a much faster carrier wave, into audio frequencies (though usually not above a few kHz).

At faster speeds you don't hear the tremolo part of the effect anymore, so the main effect becomes the harmonics.

Which harmonics depends on how the sound and the carrier wave relate. they usually don't relate directly, which implies the effect is usually dissonant.

The output contains only the modulation between the two, and won't contain the carrier or original input as they were.

For a pure input tone (and pure carrier), the output of this waveform multiplication is essentially a mix of two new waveforms, and the frequencies of these two will be the sum of the two, and the difference. (this makes somewhat more intuitive sense when you find a graphical explanation - look around until you find one that makes sense to you).

This is also why ring modulation with a very slow carrier (<10Hz or so) will sound like tremolo and not weird yet, mainlybecause only similarly-slow (sub-bass) signals will be altered much (again: sum and difference) which you barely hear, and higher ones are altered very little.

For higher frequencies, though, and for free-form sound, For free-form sound, RM generally sounds a bit dissonant, metallic, sci-fi, and/or creepy.

This decription varies with both frequencies - including the harmonics already present.

For example,

~25Hz

applied to voice (~half-kHz) mostly seems to be adding a bit of bass,

applied to piano in mid-range (~1-2kHz) seems like a woblly detuning.

100Hz+

applied to a higher-pithed voice sounds Dalek/Cyberman,

applied to piano seems like a less-wobbly version of the same detuning.

There are some combinations that are less obvious, because they are less dissonant.

For example the application to just drums, the combination with FM,

Analog, digital, and design notes

Ring modulators are most associated with analog circuits (two transformers, four diodes), in part because of the very name: 'ring' refers to how those diodes are set up.

This happens to be the simplest circuit. The diodes effectively react to whether the carrier wave is in its positive and negative part, effectively working as switches.

In theory we could try to make one with just diodes(verify), but the transformers help make this a balanced modulator, which basically means we have the carrier be canceled out on the output.

This design is also nonlinear (around 0V) because of the diode behaviour, adding some distortion. This is unavoidable, though often fine as they're typically for noise making and distortion in the first place.

Still, be aware that there are cleaner implementations of ring modulation than that.

Say for a moment you indeed looked up a video on the what it does with waveforms. You'll remember that in terms of the waveform, is essentially direct multiplication of the samples at the same point in time.

That sounds like something that you could easily do digitally, if you had all the waveforms around. And this how the C64 SID chip ring modulation works, because it has its own oscillators's outputs right there in still-digital form.

This is cleaner than the analog form, because the near-zero diode dropout is gone.

In theory you could tune ring modulation to track your keyboard, so that it has a more controlled dirty detune-ish effect, at least to monophonic playing.

Optional features:

• Wet/dry mix

• an oscillator (LFO to a few kHz)

to make external carrier optional rather than mandatory

• a lowpass after

to control the sound a bit.

you could make this relate to the internal oscillator's frequency

Diode ring modulator

You often want the option of amplification and/or attenuation before it, because it's passive,

Design considerations:

• germanium diodes are recommended due to their lower forward drop.

this matters slightly less in modular voltage levels, than in typical line/consumer levels

(schottky?(verify))

• matching diode forward drop makes for a more balanced modulation

again, not the most crucial if it's a noisemaker

• transformer ratio - generally 1:1.

might a "1:more and back" setup help with the diode drop issue?

• gain on input, attenuation on output

same potential reduction of diode noise

• tranformer impedance doesn't matter hugely

it affects

• output buffer

not really necessary (why?

can be useful, though assuming a transformer with moderate impendance (preferably a few kOhm)

• ac coupling on the input

often not necessary, but

IC ring modulator

You can get similar results with ICs like AD633 ($10), which is a cleaner variant because it avoids the diode drop.

http://musicthing.co.uk/modular/building-an-ad633-ring-modulator-module/

https://musicthing.co.uk/building-an-ad633-ring-modulator-module/

Unsorted

https://www.cgs.synth.net/modules/cgsrr.html

https://www.muffwiggler.com/forum/viewtopic.php?t=151265&sid=5e3369af220472d072772a2381d4935d

https://web.archive.org/web/20080518215055/http://mysite.du.edu/~etuttle/electron/elect61.htm

http://webaudio.prototyping.bbc.co.uk/ring-modulator/

https://en.wikipedia.org/wiki/Ring_modulation

https://en.wikibooks.org/wiki/Sound_Synthesis_Theory/Modulation_Synthesis

https://electronicspost.com/ring-modulator-for-the-double-sideband-suppressed-carrier-generation/

https://www.youtube.com/watch?v=6mzjrhCguc4

http://webaudio.prototyping.bbc.co.uk/ring-modulator/

http://www.keyboardmag.com/analog/1318/on-synthesizers-amplitude-and-ring-modulation/57208

Theory side: https://www.youtube.com/watch?v=junuEwmQVQ8

Simpler-synth building blocks - modules, if you will

VCO

An oscillator makes sounds of a particular type (wave shape).

Most are very very regular, and constantly on (requiring a VCA to be usable)

There are a few exceptions, e.g. the self-dampening bridge-t oscillation as used in drum machines.

A Voltage-controlled oscillator (VCO) means you can control its frequency, often via a keyboard, or via CV (in modular). This implies a particular way of mapping voltages to pitch, and fairly precisely.

LFO

A Low-Frequency Oscillator (LFO) is an oscillator that is very slow, typically used for control of parameters (e.g. via CV), rather than for its sound.

These aren't always voltage controlled, and when they are, not necessarily in a way that relates to pictch.

Some VCOs are capable of being LFOs (often with a switch that lowers the frequency a few octaves).

Sub-oscillator

A sub-oscillator is in essence a frequency divider, which was a way to cheaply add a one-octave-lower sound to an oscillator's sound, making it sound beefier.

They are typically square wave, because of the very simple implementation (flip-flop when analog, can be simple IC in digital).

These are probably more common in keyboard synths than in modular designs.

https://www.reddit.com/r/synthesizers/comments/1e0zwa/question_what_exactly_is_a_suboscillator/

VCA

A Voltage-Controlled Amplifier/Attenuator (VCA)

• takes an input signal (often audio, sometimes CV if you like to modulate CV)
• takes a CV that should control the amplitude

(often from an envelope generator)

• produces the input signal amplified (usually mostly dampened) according to that amplitude CV

The simplest application is to combine envelope (generated from an ADSR module based on a "you played a note" gate) to a continuously-generated tone.

But they are much wider tools than that. The "you can never have too many VCAs" argument is basically that they are the most generically useful tool you have.

While applied to audio it's a volume thing, applied to CV it's modulation of any parameter, which is directly applicable as things like like compressors, side-chaining, AM synth, and one more level of them gets a lot more interesting - this is a and in general can help insert subtle variation that make sounds, patterns, and modulation more interesting.

VCAs for synths often provide both linear response

Envelope generator

Puts out CV of a particular slow-changing shape when triggered.

Typically used to controls VCA on audio to vary the amplitude, and/or VCF on audio to vary the harmonic structure, two major, sometimes-underestiamted parts of timbre.

The best known envelope is ADSR, controlling:

• Attack: time to rise to max level
• Decay: time to fall to sustain level
• Sustain: level of sustain
• Release: time to fall to nothing

Typically controlled via a gate:

• Once high, it will do the steps to sustain.
• Once low, it will do the release

Variants without sustain can be controlled by a trigger, or by the rising edge of a gate

And many variants exists, for example settling more explicitly what happens when the sustain level is never reached, different stages, different-shaped falloff, arbitrary shapes (mostly in software)

VCF

The Voltage-Controlled Filter (VCF) is one of the more central pieces of hardware that is also quite varied.

And one of the first where a lot of personal taste can come in.

Resonance

Designs with resonance mean that at the knee frequency you get a slight boost, net effect being emphasis.

VCFs set at high resonance self-oscillate and produce someting much like a sine way, which is how they sometimes double as a sine-wave source, or as modulating CV signals.

Q factor

Terminology:

• Shelving filter - those that have equal response before and after their transition.

this tends to sound more natural than a narrow boost

lowpass and highpass are often selving {{{1}}}

• equalizer - often either wide filters to relatively adjust e.g. low, mid, and high

or, when there's a whole bunch of slider pots, it's often called a graphical equalizer, referring to how they resemble the frequency graph

Sequencer

Playing melodies without using fingers means getting pitch CVs and 'play a note' gates/triggers from a module.

Which is what sequencers do.