Predominantly Erroneous (Exohedron nonsense blog)

Musings on Electronic Music Production part 1: Sound Design part 1/?

Imagine you're a little pig and you want to build a house. So you think about what materials you might want to build your house out of. You have straw, which is nice and breezy and plentiful easy to build with, or you have sticks, which are a little harder to build with but a little sturdier, or you bricks, which are hardy but resource-intensive and also yield quite a different aesthetic than a house made of straw.
The point is that you don't really have all that many choices for building materials, or at least, not that many that would hold up against city building regulations, and really you're more concerned about how big a kitchen you can make and how many guest bedrooms you can fit in.

But suppose that suddenly a magical machine appears that will produce all sorts of materials, all of them allowed by city ordinance, and now you're not limited to straw, sticks and bricks. You can have gingerbread, or marble, or concrete, or even materials that don't exist, like whatever superhero costumes are made of. You tell the machine what properties you want the material to have, and it spits it out.
Now you have choices, and maybe, maybe those choices will affect what kind of house you have, in ways that are different bedroom arrangement.
And sure, you could stay with just straw, or just sticks, or just bricks, and focus on where to place your ovens. But you could also make a staircase that glows only on the full moon, and wouldn't that be sick? If you could do that, you might end up building more staircases.
Or you can make something that looks like straw but is as strong as brick.
Or maybe a front door that releases wolf-repellent when someone blows on it.

Anyway, the point is that you can suddenly build your house using more than just the traditional materials, and now you have to make decisions about those materials.


Okay, ditching the metaphor for a moment.

Timbre is the non-pitch information that we get when someone plays a note on an instrument. Even if they're playing the same note, the sounds that you get from a violin or an accordion or a tuba are recognizably different; you'd listen to the notes and be able to say "that's a violin" or "that's an accordion" or "that's too loud to play in a broom closet, you idiot!".
When composing, you pick out a succession of notes and times for when those notes should occur, and depending, you might decide what instrument should play those notes. You might decide that the piece would sound best on a piano, so you specify "piano concerto".
And occasionally you might stumble across a new instrument, and think "rather than a piano, maybe the piece would sound better using mayonnaise."
But you're still probably selecting from a list of things, rather than deciding features and building a new instrument.
Oftentimes, the word "arranging" is used to talk about instrument selection, as distinct from "composition" which is the determination of what notes occur when.

In electronic music, you have a magical machine that will produce all sorts of timbres. It's covered in dials and knobs and buttons and sliders and the documentation generally isn't great, but if you keep at it, you can get some really new, interesting stuff. You can make a timbre that's somewhere between a triangle and a bagpipe, or between a harp and a laser gun. Or maybe an instrument that starts off sounding like a harp but morphs into a laser gun if you hold a note for long enough.
But now in addition to simply composing, you've been sucked into the role of arranging and beyond, since you can now create new instruments instead of just picking from a list. That expansion of arrangement into engineering is called "sound design", and is one of the things that distinguishes composing for electronic music from composing for "acoustic" music.

 

This is completely fascinating to me. You’ve only just started but I’m already learning new things.

Spoiler

I fell completely in love with electronic music within the first thirty seconds of when I first pressed play on Infected Mushroom’s album Classical Mushroom. I had a few measures of percussion at the start of Bust A Move before I became suddenly, dreadfully aware that I was going to have to get up from the computer desk and dance, and somehow I would just have to manage not to pull the headphone cord. I guess I did okay because I don’t remember headphone disaster being a major part of what followed. I think it was a pretty long cord. Then they had the sheer nerve to end the album with Dancing With Kadafi. (I think I remember that Kadafi was someone’s cat, which raises far more adorable questions than any it answers.) Which I did dance to, because that was also then my new favorite thing I’d ever heard. There was a lot of awkward spatially constricted dancing. (I didn’t realize until I saw a video of a live performance filmed from an angle above the stage that the reason the piano in Bust A Move sounds so hauntingly authentic is that the guy with the classical piano skills, who wrote that bit iirc, is playing the first iteration on an electronic keyboard before hitting what I assume is a loop button. It all makes perfect sense.)

My background has a bit of flute playing; at one point I could read music. But mostly I was a classically trained ballet dancer. I never cared enough to get professional, but I did train for nine years. Most of what I know about music is a consequence of learning how to dance to it. You don’t have to have any concrete idea about structure to do that, but it works way better if you can anticipate it. Barre exercises are great for getting a feel for it without knowing, because they often happen in fours. If it’s a waltz or something then it’s still usually four groups of three. That’s the backbone of how people count music out, pretty much regardless of what they’re doing. The flute sheet music probably helped my understanding a lot by teaching me about the idea of measures.

I love electronic music because of what I can feel it’s doing to that instinct. The beat tends to be faster, but the changes tend to become glacial. It often uses your memory as a live loop device. You can dance with the music that’s currently happening using your memory of a line that just dropped out. I don’t typically use classical ballet vocabulary because ballet assumes a great deal more space than one usually has in a club or near a computer desk, and worlds more constant training than I have any interest in doing. But I can use the foundations from ballet, and as long as the foundations of music are present, I can dance.

A lot of what dance can offer to music is interpretation. (And sure, you can make synthetic dance and then you have exactly the same problems you face with synthetic instruments where all the expression has to be deliberate or it won’t exist.) I really like being a person dancing to electronic music. I am the dropped and stuttered beats and the notes drawn out a little further than was strictly written. I elevate the rise of the melody and I land the percussive attacks. So does everyone who dances, really. You don’t have to be a good dancer, you just have to decide to move, and then realize you can also decide when not to move. Electronic music tends to give you plenty of time to work it out, which I appreciate.

You can see people in clubs sometimes hesitate a bit during the first instance of whatever we’ll be hearing at least three more times before the next big change, then dive right back in with sureness. Even if they don’t realize they know, they still definitely know. That’s something that you’re probably far less likely to get from classical music, unless maybe you’re doing a lot of ballet warmups. It’s just not out there in such an accessible way that encourages people to explore and react freely. It’s a shame that people don’t seem to see that as a problem with classical music instead of with all the other music. Classical Mushroom the album probably exists in response to people being elitist about “real” music as opposed to this psytrance stuff.

Maybe if everyone took more time to consider the acoustic potential of mayonnaise in a variety of closets. I’m not saying it would solve anything, but it is both vaguely alarming and extremely funny. Especially if you imagine someone complaining that the mayonnaise is too loud for that closet.

 

Spoiler: the post on "how to play the same note over and over again for thirty minutes" is going to be mainly about trance, because that cluster of genres, in particular the psychedelic cluster and its predecessors, are prime examples of how timbre can be used the way melody can.

 

Also, since you said that you like electronic music because you can predict the rhythms, I'm probably going to be making a post about a cluster of genres that are precisely anti-that, the drillnbass/IDM/breakcore cluster. They usually have a notion of a time signature, but each bar usually has a completely different rhythm, so you either have to have the track memorized or be purely reactive.
Also they don't really allow much timing-based interpretation in terms of playing, because they're going so fast that even a few milliseconds of timing sloppiness would cause notes to overlap.
But that will be later.

 

The problem with a lot of science fiction plants is that too many of them are green. Plants here are green because of the properties of the sunlight that we get; on a planet with a different sun, a different atmosphere, the sweet-spot for photosynthesis might well be somewhere else. And furthermore, why would we expect the plant-shaped things to perform photosynthesis? There are plenty of sessile, vaguely-plant-shaped organisms here on planet Earth that don't bother with photosynthesis.

 

Musings about Electronic Music: Sound Design part 2: Sources of Sound

When people say "electronic music" you probably at least think a little bit about beeps and boops and those fun little noises that computers make according to movies and TV shows created by people who have never (or are pretending that they have never) actually met a computer.
And those are important! Well, in the sense that those kinds of beeps and boops are emblematic of electronic sound synthesis.

There are roughly two basic sources of what we might call "raw" sound for electronic music: oscillators and samples.
A sample is a recording of some sound, maybe a snippet of music or a birdsong or a particularly interesting door creak. In legal circles, sampling can also mean using melodies composed by someone else. Like with the notion of "fanfiction", what exactly constitutes sampling and the moral and economic implications thereof gets debated by people with all sorts of backgrounds and motivations, so I'm not going to get into the wider picture.
A lot of early electronic (or electronic-adjacent) music was based upon stitching samples together, sometimes literally stitching together bits of magnetic tape with sound recorded on it.
Also there are a lot of sounds that we don't know how to generate from scratch electronically; the human voice is probably the biggest challenge, considering how complex it is, but even sounds like acoustic guitars are still fairly complicated, so guitar parts are often sampled. In fact, you can find massive libraries of sampled acoustic instruments, with samples for different pitches and different accents and such. Hans Zimmer's film-scoring style evolved out of his use of samples,

We contrast this with oscillators. An oscillator is a little program that, well, oscillates: it puts out values that bounce up and down in some pattern that usually repeats. If it bounces up and down rapidly enough, we perceive the resulting air vibrations as sound, and if it bounces up and down with a regular pattern, we perceive the sound as pitches.
The timbre of the sound depends on the function. The simplest, for some definitions of simplest, is the sine wave, which varies up and down smoothly. You hear whatever the frequency of the wave is, and that's it. It is super dull.
More complicated functions that often come from oscillators are square waves, which sit at 0 for a bit, jump to full and then stays there for a bit, and then drop back down, and triangle waves, which slide up from 0 to full and then slide back down, with the change between sliding up and sliding down being abrupt rather than smooth, and sawtooth waves, which slide up from 0 to full and then drop down immediately.
These waves have a bit more character to them; in some sense, they're actually made of lots of sine waves stacked on top of each other; for each of these less-basic waves there's a sine wave that's going at the same frequency, but then there are higher-frequency sine waves as well called "harmonics" which help create the timbre. Sawtooths can sound a bit like a whine, like a mosquito, while square waves are stereotypical computer sound.
Theoretically you can use any shape as an oscillator pattern. The distinction I'm drawing between oscillators and samples is that a sample is fired off once per note, while an oscillator goes through the pattern over and over again until you tell the note to turn off. So you could, for instance, have the entire soundtrack of the Bee Movie as your oscillator pattern, and in that case if you hold a note for long enough it would repeat, whereas a sample-player would simply play the soundtrack once per note and then wait for you to press another note.
Oscillators can also be set to produce noise, which in this case means that the bouncing up and down is random (or at least, pseudorandom) rather than following a repeating pattern. Noise can also be viewed as layered sine waves, and again there are different timbres based on which waves you have present.

The fun part, in my opinion, comes from what happens after you've gotten your sound, either via sample or oscillator. While with, say, a piano, once you've gotten the piano to make the sounds corresponding to the notes you want, you usually just record the sound that comes out of the piano, for electronic music generating a wave is only the first step. The next step is the effects, modifying the sound you have, and that's where the real power of electronic sound design comes from. But that's another post.

 

I like how she kept saying "autograph" instead of "signature".

 

So in the game 2048, you have a 4-by-4 grid with some tiles on it, each tile bearing a number. You can swipe up, left, down or right, and if any tiles can move in that direction, they will move, as far in that direction as they can go. A tile can move on to an empty space on the grid, or if it meets another tile with the same number on it the two tiles will combine into one tile whose number is the sum, i.e. twice either of the original tiles.
If any tile manages to move or combine, the game dumps a new tile randomly onto one of the empty spaces on the grid; if nothing can move or combine, you lose.
The goal of the game is to reach, via this combining method, a tile with the number 2048 on it, but you can ask to go further. This sets off what is called "endless mode", but there is an end; eventually the board will fill up regardless.
After all, to reach 2048, you must first reach 1024 not once, but twice, and until you reach that second 1024 the first one will take up board space. To reach that second 1024, you must first reach 512, not once, but twice, and until you reach that second 512 the first one will take up board space. And so on, down to the smallest tile.
The game drops tiles marked with 2s, so the filled board has a 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768 and 65536, where 65536 = 2^16, and looks like this:

and nothing at all can move.

Spoiler

Except instead of a 2, sometimes it drops a 4 instead

at which point you can collapse everything and go even further beyond:

 

If you happen to have any recommendations for any of those genres, I’d be very interested! I can’t immediately think of much besides Aphex Twin where I noticed that there was a goal other than “don’t do the normal thing”, which is valid but not necessarily something I’d think to seek out unless I’m looking to throw down with the song in some sort of dance battle, which I’m usually not. (Though I’ll fight Windowlicker any day because by the time it gets to the part that hits hardest, the song sounds like it is beginning to crash down around you in pieces, which is glorious.) It’s entirely possible I’ve heard great stuff and either didn’t know the genre or didn’t have the foundational knowledge to appreciate what it was trying to do. Or both.

 

Some names, roughly in order of loudness/aggressiveness of the music:
Boards of Canada, Autechre, Hrvatski, Duran Duran Duran, Squarepusher, Venetian Snares, Drumcorps

To be honest, a lot of braindance/IDM is deliberately anti-pop, anti-dance-music, a kind of pseudo-intellectual snobbery that, like a lot of such snobbery, is buried under enough layers of irony in the Striderian sense that there's not always all that much actually going on under the hood. Like a lot of art made as deliberate rejections of the conventions of popular culture, most of it is honestly kind of bland and uninteresting.

 

Thank you! I’ll check them out!

If I’m dancing anything remotely rooted in ballet to electronic music, I figure I have preemptively forfeited my right to complain too loudly about snobbery. I admit to thinking that some of what I have heard in the past reminded me of a DJ who was suddenly realizing they were very ambivalent about their entire catalogue, but I try not to become so amused by myself that I forget that someone probably put a lot of effort into it. They may well be working on making their grand vision possible, and this isn’t it yet, or I’m just not the intended audience and they haven’t achieved their ideals while also transcending the boundaries of personal taste. It may not all be my cup of tea, but if you’re gonna deconstruct the conventions of music itself, you probably gotta start somewhere. Most of everything is kind of bland anyway. We don’t all have to be Aphex Twin, regardless of what the music video may seem to imply.

 

why am I here listening to people talk about math when I could be looking at cows?

 

On the occasions that I think about what I might want in a life partner, I end up with basically one main criterion: I want someone smarter than me. I want to spend my life with someone who can teach me something every day. Hopefully mathematics, but I guess I wouldn't mind other stuff.

 

Musings On Electronic Music Production: Sound Design: Part 3: Effects part 1/?: Volume, Filtering, and Envelopes

Imagine that you get set up to take a selfie and you figure out the Objectively Correct pose, but decide that no, this isn't quite the right flavor of serious to send to your boss to let them know that you're actually sick and not just skipping work to binge-watch Hell's Kitchen. Again.
So you fire up Photoshop and scroll through the options until you find it, subtle but meaningful, that will leave most of the picture completely unchanged but will add that crucial detail: Gaussian Blur.
Perfect. Send.

One of the big draws of electronic music is the ability to alter sound that has already been produced. For most acoustic instruments, once you've gotten the sound you don't really do much to it; you press a piano key, a sound comes out of the piano, and that's it, that's the sound that you want your audience to hear.
But the advent of electronics allowed for post-generation modification. Techniques and technologies originally developed for dealing with radio transmission got turned into ways to shape and distort recorded or generated sound for musical purposes. Just as photographs often get Photoshopped, either with actual Photoshop or with any number of alternatives, music these days often undergoes a large amount of processing after leaving the instrument.

There are a number of families of effects, all of them doing things to the sound wave that will eventually reach your ears, but presented differently. There are those that talk about volume, and those that talk about frequencies, and those that talk about wave-shapes, and those that do more exotic things.

Volume is the simplest: these effects change the volume of the wave, usually either in a pattern or in response to something. For example, you can take a sample and mute parts of it to introduce new rhythmic elements (or to censor it in humorous ways) like stuttering.
A common volume tool is called "compression", which tries to deal with two of the big problems in electronic music: 1): If you try to push too loud a signal through your wires/speaker, it ends up sounding bad because of the limits of the physical objects, and 2): below the pain threshold, louder things often sound "better" than softer things.
A compressor tries to solve this problem by pretending to be a gremlin with a volume knob, such that whenever the music would get too loud, it yanks the volume knob down, and when the music gets too soft, it pushes the volume knob up. The stuff that would be too loud gets softened, and the stuff that would be too soft gets louder.
This isn't always a good thing; to paraphrase a supervillain, when everything is loud, nothing is loud. A compressor compresses the volume range of the music, and people don't recognize absolute volume as much as relative volume; you're supposed to shout the chorus and mumble the verses, and that does't quite work as well if they're the same volume.
Even worse, it can cause the sound to degrade, as was seen during the Volume Wars (I am not making this name up), where produces kept trying to push the compressors further and further to wring every last bit of volume out of the music without regard to quality. Remember that music is just a wave going up and down, and if you compress it too hard you just get ups without any downs. This culminated in the Death Magnetic album by Metallica, released in 2008 to much criticism because it was basically mush, like the metal equivalent to trying to talk through a mouthful of peanut butter. The snare drums are just bursts of static and I have no idea what's going on with the low end.
But it's not all doom and gloom for compressors. A trick used in a lot of dance music called "sidechain compression". Here, instead of pulling the volume down when the music as a whole gets loud, a particular instrument, usually the kick drum, is used as a signal to pull everything else down. So the kick comes through loud and clear since everything else is made quieter, and then between the kicks everything else rushes back in like the tide on a planet that spins at 250 days per minute. If you've listened to modern electronic music, you've almost certainly heard this effect to some extent. Some genres of music have even built themselves around this effect, for instance early Daft Punk and the rest of the French House scene circa the turn of the millennium, where the compressor is set so that there's a "sucking" or "pumping" sound to the volume changes.

Another type of effect acts on frequencies; rather than affecting the volume of the entire sound, it changes the volume of the various harmonics of the sound. The simplest example is a filter, which has a "pass" region and a "cut" region; frequencies in the pass region go through unchanged, while frequencies in the cut region get softened or removed entirely.
For instance, a high-pass filter will remove all of the frequencies below a certain threshold. This is useful for cleaning up sounds, which often include frequencies that humans simply can't hear but that, if loud enough, will cause the music to sound muddy or too thick. Conversely, a low-pass filter will remove everything above a certain threshold; again this can be used to remove stuff that humans can't hear, but can also be good for things like bass lines that are supposed to occupy the low region but you don't want interfering with the higher-pitch stuff.
When you've got a lot of instruments playing at the same time, you really need a deft hand with volume and filter controls because there's only so much sonic space. Setting the relative volumes so that the instruments that should be front-and-center are louder and more extraneous instruments are in the back gets you pretty far, but you can get more in if you carefully filter the instruments so that only the most important frequencies of each element get through; if the instruments aren't competing for the same frequencies, you can distinguish them better.
This is why even acoustic bands get electronically processed these days when they do studio recordings, because if you've got too many sounds going at once then you've got to start cutting into each sound to get them to all fit. Even just two guitars might need to be separated via cutting their frequencies so that they occupy separate parts of the sonic space.

Finally, I want to talk about envelopes. An envelope attached to a parameter says "when the note starts playing, I set the parameter to this value, and then to this value, and then to this value, and when the note stops I set the parameter to this value".
Usually, we have an attack portion, that goes from 0 to max, a decay portion, that goes from max to the sustain level, a sustain portion that holds the sustain level until the note ends, and release portion, which slides back down to 0.
Consider a violin, When you draw the bow across a string slowly, the sound doesn't reach max volume immediately; rather it builds up to it as you draw the bow across. That's the attack, taking a moment to get to full volume. Then as you continue to draw the bow across the string the sound stays roughly the same volume, meaning that the sustain level is roughly the full volume, and when you remove the bow the sound takes a bit to fade out, which is the release. Compare with smacking a cymbal with a stick, which gets real loud real fast, and then fades very slowly: short attack, really long release, unless you grab it in which case the sound cuts off almost instantly, i.e. short release.
If you pluck a guitar string, again the volume goes to max pretty quickly and then fades, but there is also something going on with the frequencies. It's hard to tell from listening to a guitar, but if you try to reproduce the sound electronically, you find that you also need an envelope on a low-pass filter, where the threshold starts low, jumps up instantly when the note starts, but then also drops really quickly afterward. That's how you get a pluck sound as opposed to the sustained harmonics of, say, a piano key being held down.
A lot of synthesizers these days come with volume and filter envelopes in addition to the sound generators because they're so necessary to sculpt sound. Before envelopes, all electronic sounds were very uniform, very flat: full volume immediately upon the note starting, complete silence once the note stopped, and no timbral development across the note. Just BEEEEEEEEP. Now we can get more interesting sounds that can evolve over the course of a short note, giving us pluck sounds and bowed-string sounds and piano sounds, at least approximately.
You can also stick envelopes on pitch; that's how electronic kick drums and bass drums are often produced, including the famous 808 bass drum that you hear so much in hip-hop. You get a "drum" sound by start with a high pitch and dropping it super fast to a low pitch; if you do it slowly it sounds like an old computer turning off, but if you do it fast then it sounds like a sci-fi pew, and if you do it really fast then you get an approximation of a kick drum. In the parlance of envelopes that's instant attack to high pitch, basically instant decay to low pitch, and then if you're doing hip-hop you set the sustain level to something low enough to replace a bass line but still high enough to be audible to humans (or at least, high enough that the subwoofer will make your chest rumble rather than completely pass you by).

 

You get at least some of these in other branches of digital signal processing. Probably a lot, really, but I wouldn’t be familiar with all of them. Mostly some image processing.

A low pass filter on an image can remove noise like dust or specks because it’s removing things with a lot of sudden change between pixels. That counts as high frequency. Though light is also waves, it’s generally more useful to talk about change across an image and treat the wavelengths in any pixel as static values. You have exactly three possible wavelengths to deal with in an LED pixel and there’s nothing you can do about it without commissioning your own digital display, so it would be a very repetitive conversation. So a low pass removes sudden changes across the image, rather than colors sufficiently close to violet.

A high pass can be added back on top of the original to sharpen an image, because it’s making important changes bigger. You’d want to check your minimum and maximum values if you’re adding, and then either rescale all the numbers to fit in your color range, or set every value that’s too big to your max value and everything negative to your minimum. You may not know how an image with out-of-range values will be handled. Mud is the good result. Often you get a black rectangle because the program was too smart to try to display an improperly formatted image.

I can’t help but think Metallica ought to have done the same thing to their album cover that they did to the contents. The Mostly White Album With Some Artifacty Dark Bits.

 

Musings on Electronic Music Production: Sound Design: Part 1.1/?: Filter Resonance

One thing I forgot to mention about filters is that the transition between the pass and cut region isn't quite smooth, for various reasons. At the threshold frequency, instead of going from "pass through unchanged" to "lower the volume", there's actually usually a peak, a small region of frequencies that actually get louder. This was originally due to the way that filters were implemented in hardware, using mathematical and mechanical techniques that led to this so-called "resonance" emphasizing the threshold frequency, but now it's done deliberately as part of the sound design.
If you're using frequencies to carve up the sonic space, then you usually don't want a high resonance level, since that will cause the instrument to take up more space near the threshold. But if you're using a filter with an envelope to do timbre design, then setting a high resonance will emphasize the change in the threshold as the envelope plays out. The result is kind of like a pitch change, as which frequency is the loudest changes over time, but since the frequencies are all "in harmony" with the base frequency of the note the overall effect comes off differently from a musical standpoint.

 

As I was getting in the elevator at the hotel, a mother with an almost-1-year-old baby came in beside me. It was a little late and the baby was obviously getting cranky, but the mother had a brilliant idea: give the baby a cookie!
This is not to say that the baby ate the cookie; he wasn't up to solid food yet and the cookie was a treat his mother bought for herself. But having the cookie to hold intrigued him enough to be quiet at least up to their floor.

 

Inexperienced exorcist: so it turns out that "ghost pepper" means something other than what I thought it did.

 

One of the problems with being "just skin and bones" is that, as advertised, I have very little padding, so when I bang a limb against, say, a table, or a door frame, or my subwoofer, it goes straight to the bone and the stinging lingers for at least a day and a half.

 

If I fall asleep while sitting I often somehow manage to do so holding up my head with my hand not anywhere on my chin or my cheek or anything like that, but pressing against my voicebox. And so I doze for a bit and then wake myself up with a noise that's like a grumbling snort or something, which I have never managed to reproduce while awake.