2

I'm looking at synths these days. A subtractive-synthesis guitar pedal with the ability to create new waveforms from scratch with a couple of oscillators may be in my future, something like a BOSS SY-300 or a Meris Enzo.

But there are specific sounds that I'm looking to recreate. Even though I'm dealing with subtractive and not additive synthesis.

How can I learn how to create synth waveforms that correspond to particular sounds? Are there devices or software that can take a waveform and break it down to a particular synth configuration? Are there media or software that I should get to help me analyze the form?

  • @YourUncleBob yes but it doesn't allow creation or major modification of new sounds, "just" 121 sounds (many of them useless) with a couple of knob twiddles. – pro Jul 23 at 22:31
  • So you want a synth sound recipe catalog? (Something like the synth version of the 'Find a Font' website?) – Randy Zeitman Jul 26 at 21:16
6

I think most people learn by mucking around with synths and seeing what sounds come out; as you do this, you learn to associate certain synth architectures (and certain parameter settings) with certain sounds, and get to the point where you can make the association in reverse. (You also learn which synth architetures are more suited to certain sounds).

If you want to do this in a more directed way, Synth Secrets is a classic tutorial, though it isn't focused on any one particular synth architecture.

With all the free/cheap virtual synths available these days, it's a much better time to learn than when I was learning - getting hold of an analogue synth back then meant saving up a lot of pocket money!

You may find the answers on Learning to Patch Synthesizers - "Synth Theory" useful too.

3

There is software that will analyze an audio file and calculate how to re-synthesize it, but it is geared mostly towards additive synthesis and frequency modulation synthesis.

In the case of additive synthesis this is logical, because if you know the level of each harmonic of a sound, you can recreate it exactly with additive synthesis. The drawback is of course that this recreates just a single-cycle wave with perfect-multiple harmonics, while it is often the slight inharmonicity and the changes in timbre, pitch and amplitude over time that make a sound recognizable and interesting.

In the case of FM synthesis, software to re-synthesize existing sounds emerged primarily because FM synthesis is far from intuitive, and most people found it near impossible to dial in the sounds they wanted to make on e.g. a Yamaha DX-7.

To an extent, you can use a plot of the harmonic spectrum of a sound to help you recreate it using subtractive synthesis. If you have a basic knowledge of the spectrum of the waveforms typically found on analog synths (sawtooth = all harmonics at level 1/n, square wave: only odd harmonics, pulse wave with 1/x duty cycle: all harmonics with a dip around every x-th harmonic, triangle wave: only odd harmonics but at level 1/n^2, sine wave: only first harmonic) you can look at which harmonics in the sound are the loudest, and see if they follow a pattern that you can re-create by combining, detuning and filtering the basic waveforms.

Tutorials about how to re-create certain sounds will tell you e.g. that an electric bass is best synthesized with a narrow pulse wave and low filter resonance, or that strings require multiple detuned sawtooth waves or pulse width modulation, combined with a chorus effect, etc. However, this knowledge doesn't really help you if you want to recreate the sound of a different instrument, or that strange bird-like siren you once heard in a movie. At that point, I think experience of fiddling around with a subtractive synth and trying out every setting will be much more important than any knowledge you can gather from a book or online tutorial.

If you want to explore the possibilities of subtractive synthesis to recreate certain sounds, in order to get a good idea of what kind of guitar synth setup you want to buy, my advice would be to buy any subtractive synth that has a good feature set and a knob for every feature on the front panel, and just experiment with it for a while. It needn't be expensive, vintage, real analog, stylish or from a big-name company, any synth with decent quality should be enough for learning and exploring.


I had a look look at the synths in stores now, with the following wish list of features in the back of my head:

  • Two detunable oscillators with the standard waveforms including pulse width modulation and white noise, oscillator sync, ring/amplitude modulation, and frequency/cross modulation.

  • A 24dB/oct (4-pole) resonant low-pass filter, capable of self-oscillation, with switchable key sync, plus a high-pass filter. Possibly also a 12dB/oct filter option.

  • Two four-stage ADSR envelopes with switchable key follow.

  • A low frequency oscillator with delay/fade-in and different waveform (preferably both sine and triangle).

  • An interface where every feature has its own knob or button on the front panel. I think this is crucial for an enjoyable synth experience, especially for your use case.

  • Key Velocity is a bit of a dilemma. Many classic synths didn't have it, but on the other hand it can make a sound much more alive, even if it just modulates the amplitude. With a background as a guitar player, you'll probably find a synth without velocity to give a somewhat sterile playing experience.

  • Chorus effect isn't strictly necessary, and was usually added to fatten up the sound of a single-oscillator synth, but it is useful on other synths as well and some classic synth sounds rely on it, especially strings and organ sounds. But as a guitar player you may already have a chorus effect that you can run the synth output through.

In the affordable part of the market, there are a few smaller synths with most of the features listed above, although they each have their drawbacks. The Roland Aira System-1 is a 4-voice digital synth, and it is the most fully-featured in this price range. Its main drawback is the 2-octave non-velocity keyboard with limited key movement; you'd have to try this out for yourself. (There is a rack version you could use with a keyboard of your choice, and there is a larger version, the System-8, with a better keyboard and effects, but the same garish looks).

Similar is the Korg Minilogue, which is a real analog 4-voice synth, but it lacks some features; e.g. the wave shape of the two oscillators cannot be modulated independently. It also has smaller than standard keys (again, try it out).

If you don't mind the ridiculously small form factor, then the Roland Boutique Series JP-08, a digital recreation of the legendary Jupiter-8, could also be interesting, but you'll only find these second-hand. If you combine this with a decent controller keyboard instead of the boutique series own K-25m, this would make a great introduction and learning setup.

If you don't mind the synth being monophonic, then the re-released ARP Odyssey is an analog classic, with many interesting features, and in my opinion more interesting feature-wise than a SCI Pro-One or Minimoog clone. Have I mentioned that it's a classic?

If you're interested in going semi-modular, there is the Behringer Neutron. It is analog and has an extensive modulation patch bay which allows weird and unusual things that most hard-wired synths can't do. Drawback are that it's monophonic and doesn't have its own keyboard.

(There are also the Behringer DeepMind, Roland SH-01 Gaia, and the Arturia range, but they each have drawbacks that make them less suited for learning subtractive synthesis. At the higher end of the price range are various interesting options from Dave Smith, Sequential Circuits, Novation, Nord and Access.)


If you're not just trying to decide which guitar synth to buy, and you want to dive into the synth world and invest some time and money in it, experimenting with a graphical modular synth programming environment like Reaktor (or a second-hand copy of an old version thereof) will probably teach you the most about sound synthesis. It allows you to build exactly the synthesizer or sampler or effect or ... that you want to experiment with from scratch (or modify one that someone else has made), and includes oscilloscopes and other kinds of visualization, so that you can look at the waveforms while you listen to them.

Everything you read in books or online about sound synthesis can immediately be put into practice. Whether you need a synth with ten oscillators per voice, or a combination of frequency modulation and oscillator sync, or granular synthesis, or VoSim technology, or additive synthesis, or any weird scientific stuff you read about, you can try it out without needing to own a bunch of different synths that each only do one thing.

One prerequisite however, is a slightly nerdy disposition. If you have zero interest in acoustics or maths or physics or programming, you may find that there's too much fiddling around on the computer and not enough actual playing going on.

For an example, have a look at this answer I wrote to a question here about using Reaktor. The asker wanted to experiment with Shepard tones, and I didn't really understand what he wanted to do, but while thinking about it, I started to wonder whether you could build a filter bank with constantly rising or falling filter frequencies that could make any input sound like a Shepard tone, and I built such a thing with Reaktor, and the result turned out to be quite interesting and useful.

Here's another answer with a simple Reaktor instrument that also shows what the graphic programming environment looks like; you just select a bunch of parts and patch them together with wires, there's no programming as in typing in code (although you can do that if you want to in recent versions).

There are other modular environments out there, like Max for Ableton Live, but I haven't used it, and I think Reaktor will have a less steep learning curve and be a better choice for specifically learning about subtractive synthesis.

A good alternative for software would be something like the Nord Modular (look for a second-hand version G1). It's a little more limited, and won't give you all the visualization and weird interface options, but it's a real instrument that you can unhook from the computer and play anywhere, which is fun too. (If you buy a Modular G1, let me know and I'll send you some patches I made for it, which each explore a particular aspect of subtractive synthesis.)

  • Modular synths probably are what you want for working through something like synth secrets though! – topo morto Jul 23 at 7:39
  • I’d suggest the Behringer model D over the deepmind. Or neutron. – Todd Wilcox Jul 23 at 13:31
  • @ToddWilcox I've based my suggestions on features and "one knob per feature" and having a keyboard, not sound quality or analog vs. digital (and I didn't really suggest getting the Deepmind.) I'd actually suggest the Odyssey over a Minimoog clone. But I stand by my suggestion of the System-1 for learning purposes; I couldn't have picked a better set of features myself. – Your Uncle Bob Jul 23 at 14:14
  • Arturia's Minibrute isn't bad at all, except it's mono - like most analogue synths at the time. And it's self-contained, with a two octave keyboard. – Tim Jul 23 at 15:45
  • @Tim I've tried it in a shop, but there was a disconnect between the controls and the sound that bothered me; like you'd change the envelope, and it would not affect a note that was already playing. Plus the filter is a bit unusual, and the controls have strange names like "metallize"; I don't know whether the manual describes what that actually does to the waveform. Mind you, I'd recommend it in other cases, just not for this question. – Your Uncle Bob Jul 23 at 17:09
2

I recommend studying synthesis with a modular environment where you can really tweak and listen to the plain individual components, freely connect them, and see and feel how they interact. Avoid all kinds of pre-made synths, and particularly avoid using presets of any kind. If you're interested in really learning synthesis, you want to connect all the components yourself. Listen to the oscillators, look at their outputs with a scope and a spectrograph, hear what the different waveforms sound like in various combinations and with envelopes, make filters resonate, feed signal output back to input, through delays, controlled with LFOs, all of that. That's synthesis.

Stay away from:

  • pre-made synths
  • presets for pre-made synths
  • newest features
  • anything that offers gigabytes of downloads

This is not purism or elitism, you just won't understand how it works, unless you start from elemental components. The components of synthesis are very simple, they have remained unchanged for decades, and they don't require the newest anything.

Pure Data, Max/MSP, Reaktor ... they're all great, but every one of them has a learning curve just to be able to operate in the environment. Pure Data has the advantage of being free and open source. Here's an introduction to Pure Data https://www.soundonsound.com/techniques/pure-data-introduction

At first you probably won't be able to recreate the sounds you want, but the learning steps have to be taken one by one, and there are no shortcuts.

Hardware modular systems can be great, but they're dangerous because they eat all your money, time and space, and you'll become a modular synth freak. I've seen warning examples.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.