Tuesday, January 29, 2013

Subtractive Synthesis Part 2



This part and the next part will examine "the practice": what the typical buttons, knobs, and sliders on a subtractive synth do to a sound. I hope to provide a starting point for developing your ability to hear a sound and "reverse engineer" it, and of course for designing your own patches from scratch. In this particular installment, I will briefly explain the meaning of the following terms and their relevance to patch design: oscillator, tuning / detuning, polyphony, filter, resonance, and "opening up."

Oscillators

An oscillator -- often abbreviated "Osc" and sometimes "VCO" for "Voltage-controlled Oscillator" -- generates the basic waveform(s) of a subtractive synth. The number of oscillators on a given subtractive synth is usually from around two to six. The character of a finished sound will vary drastically according to what waveform(s) your oscillator(s) is generating. Some synths allow you to choose from a large number of different waveforms for the oscillator to generate, and some even allow you to draw or load your own waveforms and play them.

Many synths give you the ability to fine-tune the pitch(es) played by the oscillators in terms of "cents." A "cent" is a tiny pitch difference of just 1/100th of a semi-tone (the musical interval from a note to its flat or sharp). Detuning refers to a process in which two or more oscillators are used to make a sound and one oscillator is pitched higher or lower than the other. This creates a sort of "vibrato" or "chorusing" effect when two or more waveforms of slightly different frequencies interact and bounce off of one another. It is the basic idea behind the famous "supersaw" sound, which in its widely-known JP-8000 incarnation used seven sawtooth waveforms detuned from one another.

Basic detuning example

The above file plays a single sawtooth waveform followed by two sawtooth waveforms detuned from one another by thirty cents.

Polyphony

Polyphony is the number of notes that a synth can play simultaneously. Many early synths were monophonic; they could only play one note at once. Most modern synths allow you the option of setting their polyphony number. In terms of softsynths, the polyphony setting is part of what determines the "upper limit" of a synth's CPU usage; a synth patch set with a relatively low polyphony number will be less likely to overload your CPU simply because it will not allow you to play enough notes at once to do so.








Filters

A filter is a part of a synth which attenuates ("cuts out") certain frequencies of the waveform generated by the synth's oscillator. Which frequencies are cut out will depend on the type of filter and the cutoff frequency of the filter.

The four most common types of filters are:

Lowpass (LP): Cuts out all the frequencies above the cutoff frequency.

Highpass (HP): Cuts out all the frequencies below the cutoff frequency -- the inverse of lowpass.

Bandpass (BP): Cuts out all the frequencies below and above a narrow "band" of frequencies.

Notch: Cuts out all the frequencies within a narrow band and lets all others through -- the inverse of bandpass.

Basic examples of what these filters do to a sound:


A raw sawtooth waveform
A sawtooth waveform with a lowpass filter
A sawtooth waveform with a highpass filter
A sawtooth waveform with a bandpass filter
A sawtooth waveform with a notch filter

The resonance setting on a filter allows you to increase the volume of the frequencies near the filter's cutoff frequency. When you increase the resonance on a filter, the sound will often seem to "squeak" or "bubble," depending on the cutoff and filter type you have set for the patch. Resonance can be manipulated to achieve some pretty neat-sounding effects, as in the following file:

Two examples of high resonance

The first part of the file shows what happens to a sound when the resonance is increased gradually. The second part is an example of what happens when a sound has a constant high resonance setting but the filter cutoff is moved up and down.

It should be noted that the term "cuts out" is not entirely accurate when talking about what filters do, since filters are never "perfect" at eliminating all the frequencies above or below a certain point. Imagine running pure noise (all frequencies playing at equal volume) through a low pass filter; a graph of the frequencies in the resulting sound would look something like this:



As you can see, the filter causes the frequencies of a sound to have a "sloping" behavior rather than a strict "on / off" one. The steepness of the slope is dictated by the strength of the filter, which is specified in terms of "dB." A 24 dB filter makes a steeper "cutoff slope" than a 12 dB one, which in turn makes a steeper slope than a 6 dB one. The steepness of the volume slope created by the filter is called the Q factor. Most modern subtractive synths allow you some degree of freedom in setting the strength of the filter.

When people talk about a filter "opening up," they mean that it is letting more frequencies through as time goes on. The term "opening up" is usually used in the context of a lowpass filter whose cutoff frequency is getting higher and higher. The sound should be very familiar to you from many trance songs, which often use filters that gradually (or suddenly) open up in order to increase the tension or energy at some point in a track. Here is an example I made today:

Two examples of a lowpass filter opening up

The first part of the file is an example of a lowpass filter gradually being opened up. The second part leaves the filter closed but then opens it suddenly at different points to create short "stabs;" even though the whole second part consists of just one note, an impression is created of a background noise and another, more intense noise bursting through to the foreground.

I hope you have enjoyed reading this part and that it has given you some ideas for sounds or at least some food for thought. If you believe that I have made any errors or glaring omissions in explanation, please tell me. In a day or two I will write and post the third part of this series.

Part Three will deal with modulation, envelopes, LFOs, and FM.

No comments:

Post a Comment