Why Two Synthesiser Sounds That Look Similar Can Sound Completely Different
There isn't one way to make a synthesiser sound — and that is exactly what makes synthesis so interesting.
If I sit at a synthesiser and select a sound labelled Warm Pad, I might reasonably expect it to sound broadly similar to the Warm Pad on another instrument.
Often it doesn't.
One may be smooth, dark and almost orchestral. Another may shimmer and move. A third might have an obviously electronic character, while another sounds remarkably like a collection of real instruments playing together.
Even more interestingly, two sounds can appear quite similar when described technically. They might have approximately the same attack, the same duration, the same pitch and even a similar frequency spectrum.
Yet play them side by side and they can sound completely different.
Why?
Part of the answer is that there are many fundamentally different ways of creating electronic sound.
Subtractive synthesis, additive synthesis, FM, wavetable synthesis, sampling and physical modelling are not simply six different brands of the same technology. They represent different ways of thinking about sound itself.
And that makes synthesis a fascinating combination of music, physics, electronics, mathematics and creativity.
Start With a Simple Challenge
Suppose I decide that I want to create a bell sound.
I know roughly what I want.
It should have:
a sharp initial strike;
a bright metallic character;
a gradual decay;
several frequencies ringing simultaneously;
perhaps a little reverberation to give it space.
That sounds straightforward.
But how should I create it?
The interesting answer is that I could use almost any of the major synthesis methods.
The result might always be recognisable as a "bell", but each method would lead me towards a different kind of bell.
That is where the experiment becomes interesting.
1. Subtractive Synthesis — Start Rich and Take Things Away
Subtractive synthesis is perhaps the classic synthesiser method.
The basic idea is beautifully simple:
start with a waveform containing lots of harmonics and remove the frequencies you do not want.
A sawtooth wave, for example, is harmonically rich.
Pass it through a low-pass filter and the higher frequencies are reduced. Open the filter and the sound becomes brighter. Close it and the sound becomes darker.
We can then control how the volume and filter change with time using envelopes.
A typical signal path might therefore be:
Oscillator -> Filter -> Amplifier -> Effects
For a bass sound, this can work extremely well.
I might start with two slightly detuned sawtooth oscillators, close the filter considerably and give the filter envelope a short initial movement.
The result could be a powerful analogue-style bass.
Trying to create a convincing bell this way is more challenging.
It can certainly be done, but subtractive synthesis naturally encourages me to think:
What frequencies should I remove?
That is quite different from some of the methods we will meet later.
2. Additive Synthesis — Build the Sound One Ingredient at a Time
Additive synthesis approaches the problem from almost the opposite direction.
Instead of beginning with a complex sound and removing frequencies, we begin with simple waves — commonly sine waves — and add frequencies together.
This connects directly with one of the most important ideas in acoustics: a complex sound can be considered as a combination of simpler frequency components.
Imagine starting with a pure sine wave.
It sounds rather plain.
Now add another sine wave at twice the frequency.
Then another at three times the frequency.
Change their relative amplitudes.
Suddenly the character begins to change.
More importantly, the individual components do not have to remain at constant volume. Each could have its own envelope.
For our bell, this becomes particularly interesting.
Real bells contain many vibrating modes, and these frequencies are not necessarily arranged in the neat harmonic pattern we associate with a string or wind instrument.
So instead of:
f, 2f, 3f, 4f...
I can deliberately introduce more complicated frequency relationships.
Now the sound begins to acquire that characteristic metallic quality.
With additive synthesis I find myself asking:
What frequencies need to be present?
Compare that with subtractive synthesis:
What frequencies need to be removed?
Same objective. Completely different way of thinking.
3. FM Synthesis — Let One Oscillator Change Another
FM synthesis can initially appear rather mysterious.
Instead of simply mixing oscillators together, one oscillator is used to modify the frequency of another.
The oscillator we hear is normally called the carrier.
The oscillator affecting it is the modulator.
Change the frequency relationship between them and the harmonic structure can change dramatically.
This is one reason FM became so famous for electric piano, metallic percussion and bell-like sounds.
A surprisingly simple arrangement can produce an extremely complicated spectrum.
For our bell experiment, FM might actually get us to something recognisably bell-like remarkably quickly.
Increase the modulation and the sound can become brighter and more metallic.
Change the frequency ratio and an entirely different set of frequencies appears.
Change how the modulation decays and we can make the initial strike bright before allowing the sound to become progressively softer.
The interesting thing is that I am no longer really asking which harmonics I should add or remove.
Instead I am asking:
What happens if this oscillator controls that oscillator?
This is why FM can feel less intuitive at first.
But it can also produce sounds that would require a much more complicated arrangement using other methods.
4. Wavetable Synthesis — Make the Waveform Itself Move
A conventional oscillator might generate a sawtooth, square, triangle or sine wave.
A wavetable synthesiser can contain many different waveforms arranged in a table.
The synthesiser can then move between them.
That word move is important.
A wavetable sound does not have to remain static.
An envelope, LFO or other control source can scan through the wavetable while a note is being played.
That means the actual waveform — and therefore its harmonic content — evolves.
This can be wonderful for pads.
Imagine creating a slow atmospheric pad.
With subtractive synthesis I might gradually open and close a filter.
With wavetable synthesis I could instead gradually change the underlying waveform itself.
The listener simply hears an evolving sound.
But underneath, two completely different processes may be producing that movement.
This is one reason two patches can look broadly similar on paper yet feel completely different when played.
5. Sampling — Why Synthesise It When You Can Record It?
There is another approach.
If I want a bell sound, why not simply record a bell?
That is essentially the starting point of sampling.
Record the sound, store it digitally and play it back from a keyboard.
At first this might appear to solve the problem completely.
But it introduces another set of questions.
Which bell?
Struck how hard?
Recorded from what distance?
With which microphone?
In which room?
And what happens when I play a note much higher or lower than the original recording?
More sophisticated sampled instruments use many recordings.
There might be separate samples for:
different notes;
different playing strengths;
different articulations;
different microphone positions;
different stages of the sound.
The result can be extraordinarily realistic.
But sampling and synthesis are not necessarily opposites.
A sample can become the starting material for further processing. It can be filtered, stretched, layered, modulated and transformed until the final sound bears very little resemblance to the original recording.
The question has changed again.
Instead of asking how to construct the sound mathematically, I can ask:
What real sound could I capture and transform?
6. Physical Modelling — Build the Instrument, Not the Recording
Physical modelling is particularly fascinating because it takes another approach altogether.
Instead of recording an instrument, we attempt to model aspects of the physical system that produces the sound.
Consider a string.
Its behaviour depends upon properties such as:
length;
tension;
mass per unit length;
damping;
where it is excited;
how it is coupled to a resonating body.
A physical model attempts to represent enough of this behaviour mathematically to generate the resulting sound.
The exciting consequence is that we can begin changing parameters that would be difficult — or impossible — to alter on a real instrument.
What would happen if the string were impossibly long?
What if its damping were almost zero?
What if a virtual pipe had dimensions that could never realistically be manufactured?
What happens when we design an instrument that obeys a mathematical model but has never physically existed?
This is one of the reasons I find physical modelling particularly interesting.
It sits at a remarkable intersection between physics and musical imagination.
We are no longer merely imitating an instrument.
We can begin designing one.
The Experiment: Make the Same Sound Six Ways
A particularly revealing exercise would be to choose one target sound and attempt to create it using all six approaches.
A bell would make an excellent example.
Subtractive bell
Start with harmonically rich oscillators, shape the spectrum with filters and use a rapidly falling envelope.
Additive bell
Combine several frequency components, including deliberately non-harmonic relationships, and allow them to decay at different rates.
FM bell
Use carrier and modulator oscillators with carefully selected ratios and a modulation envelope that decays after the initial strike.
Wavetable bell
Choose or create metallic waveforms and scan through them during the decay.
Sampled bell
Record a real bell and manipulate its pitch, envelope, filtering and effects.
Physically modelled bell
Model the behaviour of a struck resonating object and alter its virtual material, size, damping and excitation.
Then record all six.
Do they sound identical?
Almost certainly not.
But that is the point.
Which One Is Correct?
This is where synthesis becomes much more interesting than simply searching through presets.
There may be no correct method.
If I am producing a film soundtrack, perhaps realism is irrelevant.
A physically impossible bell might be much more effective than a recording of a real one.
If I am creating music for a science-fiction sequence, I might deliberately combine methods.
Perhaps I begin with a sampled bell.
I layer it with an FM sound.
I add a low subtractive synthesiser layer.
Then I introduce a slowly evolving wavetable texture underneath.
The final sound cannot really be described as belonging to one synthesis method at all.
It has become sound design.
Why Preset Names Can Be Misleading
This also explains something anyone who owns several synthesisers will have noticed.
A patch called "Strings" on one instrument may sound nothing like "Strings" on another.
That does not necessarily mean one is better.
One designer might be attempting to imitate an orchestra.
Another might deliberately be creating a synthetic string pad.
Another may be using sampled strings.
Another could use physical modelling.
Another might layer samples with synthesis.
The name describes the musical purpose, not necessarily the technology behind it.
The same applies to:
bass;
brass;
pads;
bells;
leads;
organs;
choirs;
pianos.
That is why listening matters far more than reading the patch name.
The Envelope Matters More Than Many Beginners Expect
There is another important lesson in this experiment.
A sound is not defined only by its frequency spectrum.
It also changes through time.
Imagine a piano note played backwards.
It contains essentially the same recorded frequency content, yet we immediately recognise that something is profoundly different.
The attack of a sound carries enormous information.
For our bell, a fast attack followed by a long decay tells the brain something about how the imaginary object was excited.
For a pad, we might want almost the opposite: a slow attack and slow release.
This is why envelopes are so important in synthesis.
Two patches can use similar oscillators and similar filters yet sound dramatically different because their behaviour through time is different.
Movement Is Often What Makes a Sound Interesting
One of the easiest mistakes when designing sounds is to concentrate entirely on what happens at the instant a note is played.
But many interesting sounds evolve.
A filter can move.
Pitch can drift slightly.
Oscillators can detune.
A wavetable position can change.
FM modulation can increase or decrease.
Effects can evolve.
Stereo position can move.
Even tiny variations can prevent a sustained sound from feeling lifeless.
This is particularly important when creating pads and atmospheric sounds for film and video.
A note held for ten seconds should not necessarily sound like the first 100 milliseconds repeated continuously.
Sometimes the almost imperceptible movement is what makes the sound feel alive.
VCV Rack Makes This Particularly Visible
One reason I enjoy experimenting with modular synthesis environments such as VCV Rack 2 is that many of these relationships become visible.
Instead of selecting a preset and wondering what is happening inside it, I can construct the signal path myself.
I can connect an oscillator to a filter.
Then connect an envelope to the filter.
Then use an LFO to alter another parameter.
Then introduce another oscillator.
Then perhaps use one signal to modulate another.
Each cable represents a decision.
And eventually something rather important happens.
You stop asking:
"What does this module do?"
and start asking:
"What would happen if I connected this to that?"
That is a much more creative question.
My Pergamon Makes the Comparison Even More Interesting
Working with my Wersi OAX 1000 Pergamon gives me another perspective because a modern organ/workstation environment is already capable of bringing together many different approaches to sound.
I can work with conventional instrument sounds, organ registrations, virtual instruments and software-based synthesis.
I can also explore physical modelling instruments such as Organteq and Airteq, while software such as VCV Rack allows me to move into modular synthesis.
For me, that makes the instrument far more than simply an organ.
It becomes part of a much larger sound laboratory.
And that is particularly useful when creating music for video.
Sometimes I want something recognisable.
Sometimes I want something orchestral.
Sometimes I want an electronic texture.
And sometimes I want a sound that does not obviously belong to any real instrument at all.
The technology gives me different routes to each of those destinations.
A Useful Challenge for Anyone Learning Synthesis
Rather than downloading another thousand presets, try this.
Choose one sound.
Perhaps:
Bell. Bass. Pad. Brass. Pluck. Drone.
Then attempt to create it three different ways.
Do not worry about making the versions identical.
In fact, the differences are the interesting part.
Ask:
What was easy with each method?
What was difficult?
Which controls produced the greatest change?
Which version sounds most realistic?
Which sounds most interesting?
Which would work best inside a piece of music?
That final question is particularly important.
A sound that is spectacular when played alone is not necessarily the sound that works best in a mix.
From Reproducing Sounds to Inventing Them
There is a natural progression when learning synthesis.
At first we ask:
How can I make this synthesiser sound like a bass?
Then:
How can I make it sound like a particular bass?
Eventually the question becomes:
What sound can I make that I haven't heard before?
That is where synthesis becomes truly creative.
Understanding subtractive, additive, FM, wavetable, sampling and physical modelling synthesis is not about deciding which technology is best.
It is about acquiring different ways of thinking.
One encourages us to remove frequencies.
Another asks us to add them.
Another creates complexity through modulation.
Another changes the waveform itself.
Another begins with a recording.
Another models the physics of an imaginary instrument.
And, of course, there is nothing to stop us combining them.
Conclusion — There Is More Than One Route to a Sound
Modern music technology gives us an extraordinary palette.
A musician sitting at a synthesiser today can use techniques whose origins span analogue electronics, acoustics, digital signal processing, mathematics and computer modelling.
Yet the objective remains remarkably human.
We listen.
We adjust something.
We listen again.
And eventually we decide:
Yes. That's the sound I wanted.
Or perhaps something even better happens.
We discover a sound we hadn't intended to create at all.
That is why synthesis remains so fascinating.
There isn't one way to make a synthesiser sound — and that is exactly what makes synthesis so interesting.
No comments:
Post a Comment