phase

All audio clips below were produced by snippets of Supercollider code I wrote, with some real-time adjustments.

Phase music is minimal in terms of notes played, yet still can provide a rich acoustic experience. It involves two or more instruments playing the same repeated phrase. Gradually, an instrument falls out of phase of the others, either through miniscule changes in tempo (where the offset is seemingly continuous), or an atomic beat shiftI'll only be looking at these note/beat-wise shifts in this post, but the Wikipedia page for phase music has an example of much more gradual phase shifting.. All the notes that are being played remain identical—it is only the respective phases that change over time, continually changing the character of the melody until eventually, the instruments fall back into phase. The process repeats itself.

Using the programming language/environment Supercollider, it is easy to define two instruments to be used in two patterns, playing the same melody, falling into and out of phase. First, we start with a melody, which I'll arbitrarily choose to be a diatonic major scale:

(
Pdef(\i1,
	Pbind(
	\instrument, \sea, // "sea" is predefined synth
	\dur, 0.5,
	\amp, 0.05,
	\degree, Pseq(Scale.ionian.degrees, inf),
	\atk, 0.01,
	\sus, 0.5,
));
)
Pdef(\i1).play(t);

The code below features another synth playing the same melody defined above, but this one slides through the degrees of the melody via the useful pattern Pslide. Playing \i1 and \i2 in parallel gives us exactly the phasal effect we desire.

(
Pdef(\i2, 
	Pbind(
	\instrument, \sea,
	\dur, 0.5,
	\amp, 0.05,
	\atk, 0.01,
	\sus, 0.5,
	\degree, Pslide(Scale.ionian.degrees, repeats: inf, len: Scale.ionian.degrees.size, step: 1)
));
)
Ppar([Pdef(\i1), Pdef(\i2)]).play(t); // play in parallel

A similar idea starts in the same way: one has a simple melodic phrase to be iterated over and over again. Suppose we wish to focus on one (or more) accent note(s). This accent note will be the only one that varies across iterations. We can accent it by adjusting pitch, rhythm, or both.

Through Pslide, we can make this accent note shift with each iteration. It's like saying a word with the stress on the first syllable, then on the second, then on the third, and so on. Or a sentence, with the stress traveling in units of words. As a set, the notes being played remain the same (if we keep the accent note constant). Only the location of the accent note varies.

Supercollider lets us add more complexity to the melody by introducing randomness. The code below plays a largely pre-determined melody, but the accent note is chosen at random from a pool of notes that form a predefined scale (offset by an octave). The accent note varies across iterations, so it isn't true phase music, but it's close:

(
Pdef(\a1,
	Pbind(
	\instrument, \sea,
	\dur, Pseq([0.5, 0.5, 0.25], inf),
	\amp, 0.04,
	\atk, 0.006,
	\degree, Pseq([1, 3, 4, Prand(Scale.lydian.degrees)+8, 2])
));
)
Pdef(\a1).play(t,quant: 4);

Then, we can add another instrument whose accent note travels in the opposite direction. We can do this by supplying -1 in the step argument for Pslide. We can add further rhythmic interest by playing the two patterns not perfectly in parallel, but slightly offset. If you listen closely, you can hear the higher accent notes almost coinciding on the same beat as the two patterns sync up for one period. Particularly with small \dur values and higher frequencies, playing at least two synths at small offsets in parallel produces an almost glittering effect that I really enjoy listening to.

Depending on the symmetry (or lack thereof) of the melody, it is possible to get a wide variety of interesting intervals just by shifting the phases of instruments playing identical melodies, striking a balance between repetition and evolution—even if the melody in question is very simple, e.g., a scale or its arpeggio. For this reason, phasal music is associated with the wider genre of minimal music, though the minimality lies more in the composition than the sound. Even just the example above sounds more complicated than it actually is.

The process is akin to tiling or tessellation, and doesn't require an extensive knowledge of music theory (something I don't have). There is something meditative and satisfying about listening to the evolutions of ever-changing phase overlaps, even if the actual intervals involved can be dissonant and awkward if you don't choose your melodies carefully. But the dissonance tends to not last for long, and even comes off as natural, sort of like how a lot of sounds in real life are: the wind striking metal against a flagpole, the sound of someone walking on heels, a police siren.

In this vein, here is a (biphonic?) example of two phasing synths we defined above, \a1, \a2, played on top of a backdrop of one other very-slowly phasing synth, \b (I have \a1 biased towards the left channel, and \a2 towards the right, so stereo output would be the best way to listen to this). Admittedly, \b has a very different quality due to its longer duration, sustain, and attack envelope. But at the end of the day, all of these synths play the same set of notes in nearly identical patterns. Phasing, a little bit of randomness, and minimal real-time tweaking of parameters creates a fuller, more interesting sound that constantly changes over time.