There are huge advantages to knowing synthesis techniques when it’s time to describe to a musician the ideas you have for sounds.
This is due to the language of synthesis covering the topic of describing sounds in detail so efficiently.
To make this clearer, we’d better take a tour around the various “parts” of a synthesizer, the building blocks used to generate and shape sounds. The point where we discuss envelopes of the sound kind is the main point of usefulness in communicating musical ideas to others, and we’ll get there further down this blog post. Read on and I’ll wander around the synthesizer pointing stuff out so you can start to get a feel for how it all fits together.
The oscillator is the basic tone generator of the synth, the first step in creating a sound.
Usually, in the simplest case, you would pick an oscillator, and then process it’s output at your whim using other sections of the synth, such as the filters.
There are various types of oscillator. Naturally, the first oscillators back in the 1960’s were analog circuits, built from physical electronic components like capacitors, resistors and diodes.
An oscillator in it’s most primitive form simply generates an output signal (which is actually a voltage). The signal fluctuates rapidly, and the speed of the fluctuations determines the frequency of the generated sound. This signal is called the waveform or, occasionally, the waveshape. Normally, you would find a discrete switch to select the wave you want, but in some cases there is a way to morph from one type of wave to another.
You will also find LFOs, which are low frequency oscillators. These are typically used to alter the sound of the waves produced by the full-range oscillators using a method called modulation. LFOs are extremely useful, and you can’t do much synthesis without them. They normally reside in their own discrete part of the synth’s front panel, but usually near to the oscillator section. More on LFO’s later.
There are five very common types of waveform. They are sine wave, triangle wave, sawtooth wave, square or pulse wave, and noise.
The noise waveform may or may not be generated by the oscillator section – in many synths, noise is generated by a dedicated noise section of the synth. Noise has no periodicity – it is a random source. It contains energy at ALL frequencies.
The commonest types of noise waveform are pink noise and white noise.
Pink noise has equal energy per octave, whereas white noise has equal energy per unit of bandwidth. This leaves pink noise sounding a little lower in pitch (subjectively) than the white noise does.
No matter which frequency you pick to start from with white noise, the energy between, say, 500 Hz and 600 Hz is identical to that between, say, 2000 Hz and 2100 Hz. It has equal energy per unit of bandwidth. Bandwidth is the distance between any two frequency bands. The fixed distance of 100 Hz between frequencies is the same everywhere that frequencies are that same distance apart. And whatever distance you pick, whether 10 Hz or 10,000 Hz, this is true – the given distance is equal in energy everywhere across the keyboard, wherever you start out from.
On the other hand, pink noise has equal energy per octave. Every octave contains the same amount of energy, no matter where you start from. The highest octave on the keyboard has the same energy as the lowest.
Gradually sample-playback oscillators appeared, the first of which were the Roland D-50 and the Korg M1 multi-timbral keyboard workstations, introduced in the late 1980’s. This approach was particularly helpful in broadening the range of sounds available to be processed by the rest of the synth components, because anything you could record (sample) inside the available memory could now provide sound to feed the synth’s other sections.
Sample playback instruments require a lot of samples, because they need these sounds to be workable at all velocity levels and often at all possible notes in the MIDI specification, from lowest to highest.
There are also hybrid oscillators, such as those that combine digital oscillators with analog filters. Granular synthesis, single-cycle waves, wave-sequencing and mathematical models have all found a place in providing raw sounds to feed a synth’s filters and other sections.
You will rarely hear an oscillator all by itself – other sections of the synth will go on to shape what is heard in almost all cases. Without oscillators, there would be no sound for the other sections to deal with, unless an external audio input is available on the synth, in which case you will have the option to feed external audio into these other sections.
Vocals or drums are very popular as external audio to feed a synth. This is how a lot of mangled drums and vocals from folks like The Prodigy and Nine Inch Nails came into being. However, this is not synthesis – this is effects processing, but with the synth as the processor.
A single oscillator is not very interesting by itself, but it does suit many upfront and solid sounds. If you want swirling textures such as complex, evolving chordal pad sounds, then you’ll need at least a couple of oscillators, and ideally three or four.
Sine waves are pretty tiresome on their own, making rather dull sounds until processed with filters and so on. Using a sine wave is still handy, though, even though there are no overtones in a sine wave – it’s a pure tone. If you have ever heard a so-called test tone, such as the 100 Hz, 1kHz, and 10kHz test tones used to align analog tape machines, then you have heard sine waves in the raw.
The test tone generator used for this purpose is simply a sine wave oscillator with an output level control and a way to select which frequencies to generate.
If you use a high frequency sine wave, you can blend it in to other sounds to create an overtone that rings out. More commonly, the low-frequency sine wave is very helpful as described a couple of days ago in my blog post on “SYNTHS AND SOUND DESIGN”. It works well for adding a low tone to bass drums and floor toms that beefs up their low end without revealing itself in the mix. The only issue to watch out for is phase cancellation, always a risk when combining low-frequency sounds. A quick check of the sounds summed together to a mono mix, preferably through just one single-diaphragm speaker, will reveal the extent of any problem immediately.
Typically, the sine wave is blended with other waves because the human ear likes to be interested by complexity, and the only way to get that level of interest up over time is to keep changing what is heard in subtle ways, usually involving three or four oscillators, only one of which would be producing a sine wave.
This “mix and match wave types” approach is also true of the triangle wave, a waveshape that does have a few uses of it’s own, primarily for unassuming lead and bass sounds. There are a few overtones in triangle waves, but they are not very high energy.
There’s plenty more energy in the sawtooth, which is a great wave shape for overtones. It contains all the overtones in the harmonic series, but the higher frequency overtones are weaker than the low ones. This means it is a harmonically rich wave shape, so it’s very useful for a lot of more complex sounds, such as rich, textured basses, pads and leads. Sawtooth leads are really popular, and get used a lot.
A sawtooth wave produced by an LFO is a great source for tremolo sounds, as is a square wave.
The square wave is also called the pulse wave. It’s a sound that has only odd-numbered overtones from the harmonic series, such as the first, third, fifth, seventh harmonic and so on. There are NO even-numbered harmonics in a square wave. This makes it sound hollow, thin, and reedy. A raw square wave played in the range of a clarinet can sound very similar to that instrument.
Most synths let you change the pulse width. That doesn’t change the frequency – it changes how long the pulse stays in the “up” part of the pulse cycle relative to the “down” part. The longer it’s in the “down” portion of the cycle, the thinner the sound gets. A thin pulse wave with an instant attack set in the envelope section of the synth sounds a lot like a spiky, funky clavinet sound, adding a little “Superstition” to the voodoo brew.
Oscillators may allow you to output a specific fixed frequency (like test tone generators do) or they may “track” what is played on a keyboard to determine which frequencies they should be emitting.
There is often a keyboard tracking control that allows you to set how much the pitch of the oscillator will change when you play the different notes on the keyboard.
Normally, of course, you would have each key in a chromatic sequence equal to a semitone change in pitch from the previous one. You don’t have to stick to this “piano” tuning, though. You could set the keyboard to be interpreted differently, such that the keys differ only by microtonal changes in frequency rather than a semitone at a time.
When combining oscillators to mimic acoustic instruments like guitars, it is helpful to have a component of the sound that does what the attack transient does in the real-world sound. I mean the little click of a plectrum hitting the string, that sort of thing. For instance, you can set one oscillator to produce a very short percussive transient, and set the keyboard tracking for that oscillator to only increase in pitch a cent or two as each key played on the keyboard gets higher. In this way, the click can get clickier as you play higher notes, which happens in real life with these types of instruments, but the pitch is pretty similar wherever you play on the keyboard, changing only in a very small range. You mix this sound with the other oscillator(s) producing your main body of the sound you are mimicking, and all is good.
In a similar vein, keyboard tracking can be used to reduce aliasing artifacts. This only affects digital audio. What happens is that all frequencies present that are more than half the sampling rate frequency are affected by the digital audio process and cause weird, mostly unpleasant, non-harmonic noises mixed in at lower frequencies. If you are using a sampling rate of 44.1kHz, then anything above 22.05kHz will generate these out-of-tune noises that then get mixed with the frequencies generated below 22.05kHz.
Usually, they are annoying and unwanted sound components, but aliasing does have some musical uses. When this happens, you can hear it more and more the higher you play on the keyboard, but again it will only be an issue with a digital emulation of analog sound. It’s a digital-only problem.
A likely way to minimize this problem is to change the keyboard amplitude tracking parameter (if you have one) so that the pitches causing the aliasing have less amplitude the higher you go up the keyboard.
You can sync oscillators together. One becomes the master, one the slave. When the master restarts it’s waveform cycle, it forces the slave oscillator to do the same. This means they always have the same frequency. You can’t hear the effect if you don’t have both the slave and the master turned up in the mixer section of your synth. The best sounds are from pitching the slave higher than the master. The other way around doesn’t do anything fun, really.
Pitch can be modulated in the oscillator by using a Low Frequency Oscillator (LFO). You can also modulate an oscillator using an envelope generator. It’s also easy to use MIDI pitch-bend commands (usually provided by moving the pitch bend wheel on your MIDI keyboard as desired).
The main use of the LFO is to add vibrato or tremolo to a note. There’s a MIDI mod wheel on most synths that allows you to control the amount of LFO modulation to be added. In effect, you are modulating the modulation with a second modulation control. You can also do this by having the LFO modulate pitch directly.
Alternatively, you can modulate the pitch of oscillators using an EG (envelope generator). This is where it becomes very useful for communicating ideas to other musicians.
The envelope of a sound is the behaviour over time of the sound. That means from it’s initial attack phase, through various stages such as hold, decay and sustain to the final release phase of the sound where it returns into silence after leaving whatever sustained level it had been sustaining at. This progress of the sound from start to finish gives us a very valuable way of describing ANY sounds. At all. It’s called the envelope of the sound, and, in acronymic shorthand, is often labelled as ADSR on a synth front panel, although there are various other varieties. Usually, there is a row of sliders/faders that allow you to set the levels of each stage.
ADSR stands for attack, decay, sustain, release, namely the most common parts of most sounds’ envelopes as they develop over time. There can be more stages, such as DADSR which features a decay part of the envelope directly before the attack phase begins. In that instance, you can delay the onset of the attack for a particular oscillator by using a DADSR envelope for it.
Maybe you are controlling an LFO amount so you can add vibrato to a sound. You would prefer vibrato only to appear on the longer notes. Using DADSR, you can set up the LFO to produce the vibrato effect after a short delay, so that the short notes don’t have it.
You will also encounter hold phases of the envelope, envelope followers, and break point/slope, amongst other envelope-related quirks.
In a nutshell, there is a lot that can be done working with this viewpoint of sound changing over time.
Once you get the hang of describing sounds as having these envelope stages, and recognize them in sounds quickly, you have a really helpful way to explain what you are imagining sonically. This is obviously wonderful news since sounds are often pretty hard to describe in less structured ways.
Filters are simply a means to boost or remove harmonic content from the sound the oscillator has presented to the filter.
Filters discriminate between the frequencies present in a signal and treat only those frequencies it is told to. Filters don’t generate any sound of their own.
Common filters are low-pass, high-pass, and bandpass types.
The low-pass removes high frequencies according to the settings you make, but passes audio unchanged below the selected frequency. The high pass does the opposite, letting highs pass unchanged where lows are attenuated (turned down or removed). The bandpass allows frequencies to pass only if they fall between an upper and a lower frequency limit. Again the cut-off frequency is usually a sloping behaviour with a pre-set rate rather than a discrete number.
Typically, the roll-off slopes available on a high-pass filter are going to be -6dB, -12dB, -18dB or -24dB per octave. The bigger numbers are the steeper slopes. The slope is how dramatically the signal will fall away in level to silence below the cut-off point (or corner frequency). A -6dB filter slope setting would not be as drastic a reduction in level of the lows below the cut-off in a high-pass filter.
High-pass filters are THE most useful tool available in any DAW when it comes to cleaning up the low end of your mix, and getting a really LOUD stereo master when you mix. You will use them constantly until they are second nature when you are mixing.
When you are using a filter on a synth, the filter is doing the same job as it would on a mixing console channel strip. It is sculpting the sound’s equalization (it’s tone, otherwise called it’s EQ) as desired by removing frequencies or amplifying them.
In both these filter applications (the mixing console and the synthesizer) the filters cannot add any audio. They are sound processors, not sound generators.
The audio frequencies you are targeting have to be present in some amount to begin with for the filters to be able to turn them up or down. You can remove target frequencies entirely, and usually will by losing the lows or some highs or both. What you can’t do is add a frequency with a filter that is not present in any amount to begin with. You can only turn up what is there to begin with.
In fact, it’s a ramping up and down type of behaviour centred around a so-called “corner frequency”, rather than being set at a fixed frequency where it just stops passing audio above that frequency. Usually, you do pick a set frequency, but this is normally a corner frequency setting in truth. The effect becomes more pronounced the further past the corner you go, and there are frequencies being increasingly affected as you approach the corner frequency from the “unaffected” part of the frequency range, although normally only within a semitone or three, depending on the chosen filter slope.
In most synths, each voice (each oscillator) will have it’s own filter(s) available, but it’s also true that some synths will round up their multiple oscillator signals and send them to a common filter.
Analog filters provide pleasing distortion if overloaded, and this effect is simulated with some digital filters. Sometimes, you have to turn up the oscillators in the mixer section of the synth to generate distortion at the filters.
There are a whole bunch of filter types, and they are used in synths and in mixer configurations, and are in fact simply equalization (tone shaping) tools with specific limited functions.
MODULATION AND EFFECTS
After some experimental fun, you will soon find there are some amazing things you can do with modulation.
Modulation simply means altering a sound by applying a control signal whose properties then influence the original sound. The added control signal is said to be modulating the first sound. The classic basic example is the LFO modulating a sine wave oscillator to add a sense of vibrato or tremolo to the otherwise pure and static note of a fixed frequency sine wave. This is a very Forbidden Planet kind of sound when the effect is used on a simple sine wave.
Easy to try and easy to hear. Any plug-in or synth can do this if it has a minimum of a test tone or sine wave generator (either of these is a discrete sine wave oscillator) and an LFO.
Most basic synths and plug-ins have a setting or knob for “glide” or “portamento”. This is a modulation controller too.
A signal is modulated so that it will glide a preset distance to the played note whenever you play a new note.
Modulation routing can get very complicated, if that’s your thing. You will find that amazing textures for pads filled with motion are possible with complex mod routing. You can modulate the modulation and then modulate the result, until you run out of modulators.
Plenty of experts have loaded all this complex modulation and effects sort of stuff into their products as highly tweak-friendly presets. You insert them on your Instrument track in your fancier DAW plug-in virtual instruments such as Omnisphere or Structure or Elements, and start recording or otherwise entering MIDI on that track. Start there, have fun!
Considering the MIDI language specifically allows you to use at least a dozen types of messages as modulation controls, you can see how this can get crazily complicated to explain, but it can all be experimented with in practice quite easily using your DAW MIDI tracks and editing them. Explore how to enter data with the Pen or Pencil tool in your DAW onto a MIDI track, putting it in an appropriate Edit mode if necessary. You can use MIDI note-on or note-off velocity data, note-number, pitch-bend, clock, sys-ex, channel aftertouch or polyphonic key aftertouch.
If that’s not enough control, you can modulate with Continuous Controllers 0 through 127.
That’s not enough? Seriously? Okay, you can use RPNs and Non-RPNs as well, yet more available commands ranges.
And this is all as easy as picking them from a drop-down menu and assigning them in most DAWs today.
There are all kinds of effects that synths can do. From the classic vocoder effect to triplet delays to cavernous reverb to sparkling chorus with exciter and de-esser, you can chain and apply all sorts of these babies together.
Filters of all types are handy effects processors. It seems reasonable to suppose that the ever-popular filter sweep keeps listeners engaged on a subconscious, primal level, probably linked to the merits of avoiding dangerous predators in our ancient history. A moving sound keeps us from turning away – we pause and listen.
This is worth considering when you’re trying to keep people grooving away in a spell on a dance floor, when you think about it from the production point of view.
Modulation using square or pulse wave oscillators can bring hypnotic qualities due to their rhythmic nature, so this stuff comes up a lot in dance music, whether it’s Katy Perry’s lovely pop confections or beautiful lush trance or film score soundscapes or even the weirdest underground stuff pulsing in someone’s dark and deeply disturbed basement haven.
All this experimentation delivers spectacular results pretty fast when you need complex, swirling, engaging, evolving sounds for your film/TV targeted production or dreamy love ballad. Learning to “roll your own” sounds is very helpful in talking about sound, and if you work with other musicians to produce your songs at home, you will find this stuff exceptionally handy after a while. It will also reduce your need to use other musicians for certain things.
You can mock up whole productions by yourself, then get players in for that unmistakeable human feel. Most people prefer real instruments, and at least you can add some real cymbals quite cheaply, if nothing else is a human playing in front of a microphone. Genuine cymbals really make things sound better to most people in a subliminal way for very little expense. This is due to the fact that it’s hard to fool our ears with artificial cymbals. You can use samples and add effects, but there is something completely unique in each strike when it’s really you hitting a real cymbal in front of a microphone.
It can be just one yard sale find for twenty bucks, it’ll do. Get a stand and a stick and scoot around that thing with your stick until you know what you like about it, and there you go – adding humanity to MIDI on a budget. It’s fun. Next thing, you’ll want a whole kit! Mwuhahahaha….
Now that I’ve said that, I should deny it all. One of the greatest freedoms the synth language gives you when you get conversational skills with it is the ability to create and shape drum kits precisely or, if someone else is working with you doing that job, to describe what you need to them or to an engineer or producer. Taking charge of your drum sounds is great fun and very educational.
In no time, you can knock up hand-claps, snares, hi-hats, cymbals, any percussion you like, pretty much. Research on the internet how to make these sounds – it’s pretty easy, and there are lots of folks willing to share their discoveries and experience. Try what they suggest as settings and see what you think. Tweak it using what you’ve figured out. There are vast soundscapes you can create!
See you tomorrow for the last synth post, “FUN TO BE HAD WITH SYNTHS”.
In the post after that, it will be time to talk briefly about sampling at home, but I must have my FUN first.
Once you think about sound design with synthesizers, it’s funny how the words “synth blog” sound like something sliding down a wall and hitting the floor.