Predominantly Erroneous (Exohedron nonsense blog)

If it were me I'd probably attempt to SleepBot it, if only because that's the app I have that happens to have sound-activated recording.

 

I was going to maybe write a follow-up to my Lunar New Year post from last year, but then I ate too much food and took a nap and now it's almost the end of Lunar New Year for everyone except like Hawaii and I don't really have anything interesting to say anyway, but happy new year everybody.

 

So I can get to 131072 in 2048 if I don't get distracted by absolutely anything (and allowing myself the use of the 1-use back button) but once there, as soon as I start thinking about anything other than the game I lose instantly.

 

There's that tumblr discussion about Aragorn's claim to kingship over Gondor, and the question of whether by writing such Tolkein was endorsing monarchy as the best system of government. Curiously, said discussion, from what I have seen of it, seems to leave out the fact that Denethor's claim to rule over Gondor also came from blood, and indeed stemmed ultimately from the same blood as Aragorn, and that the implication was that Boromir would have taken up stewardship had he not died, again by virtue of blood.
Now, one could make the usual arguments that Boromir, raised from birth for the role of stewardship, would be better suited to rule than Aragorn, who spent his life ranging, cavorting with elves, fighting in disguise, loitering in seedy taverns, etc. This is, however, not an argument against monarchy per se.

 

One day I will understand sheaves. One day.

 

Musings on Electronic Music Production: Sound Design part 3: Echoes

When we think about echoes, the typical image is of shouting obscenities into the mountains and then giggling when they're repeated to us. (well, okay, as a city kid my main image is the sound of fireworks bouncing off of skyscrapers (the sound bounces, not the fireworks themselves) but still).

The time it takes to echo actually tells us a lot of information about the of the space we're in. If you're indoors or in a city, you rarely hear a pure sound; rather, you hear the sound directly from the source, but you also hear all of the reflections, and those reflections color our perception of sound.
Bats are famed for using the echo time to determine how far away things are; dolphins can do it too, and humans can as well, to various extents, in the sense that most humans can tell whether they're in a big room or a small one by how things sound, and that sense can be trained into pretty good echolocation.
For most rooms that humans occupy, the walls are nearby enough that we hear the echoes pretty closely together, so we consider it "reverberation" rather than echoing. Also, the material and shape of the reflection surfaces changes the timbre of the reflections, which can lead to fun things like anechoic chambers that don't reflect at all.

Of course, electronically these effects aren't hard to make; you just have to take the audio signal and play it a bunch of times. These effects are often called "delay", when the echoes are distinguishable from the direct sound, and "reverb" when they smear together, and often come with controls to emulate different sizes of spaces and different reflection surfaces.
It used to be that faking reverb for recordings was done with a simple mechanical device of some metal plates and springs, and of course there are now electronic emulations of plate reverbs. But we also can get pretty decent emulations of being, say, in a small club versus being in a large hall; there are even emulations of, for instance, specific churches that are famed for their reverberations. You can use reverb on specific sounds to make it sound like those sounds are coming from closer or farther away, which is great if you're actually just making everything on a computer and don't have the space to actually spread your band across a football field.

Sound travels through air at about 343 meters (1,125 feet) per second, so unless the sound is reflecting off the inside of your skull (and you can fake that, too, if you want), the echo time is going to be at least dozen milliseconds or so. But you can get some interesting effects if you really reduce the time.
If you play two signals at the same time, they interfere; the ups add together to become more up, the downs add together to become more down, and if one signal has an up and the other a down, they cancel out. The reinforcement is called "constructive interference" and the cancelling is called "destructive interference". When you play two frequencies that are close together, this interference makes it sound like there are other frequencies being played as well; this is what causes what we perceive as musical consonance or dissonance.
If you play a single signal twice, but have the second copy start very slightly after the first, the frequencies of the signal that match that delay time (i.e. multiples of the reciprocal of the delay time) will cause constructive interference, amplifying their presence, while the frequencies that don't match will cause destructive interference, reducing them. We get what's called a "comb filter" since the picture of what happens to the frequencies looks like a comb, unlike a highpass or lowpass filter where it looks like a hill or a cliff. Since the frequencies being amplified are all multiples of a single frequency, the set of amplified frequencies are in harmony.
Most of the time we don't use simple comb filtering; rather, we have the delay time go up and down over time, which causes the frequencies being amplified to vary over time. This leads to effects called"flanging", named for the original production method which used tape reels with someone pressing a finger to the rim (i.e. the flange) of one of the reels to slow it down.
If you screw around with the delayed copy, like using some regular filters or whatnot, you get a more general class of effects of "phasing". In this case the amplified frequencies aren't necessary in harmony, leading to a different, somewhat more dissonant sound.

When people sing in close harmony, the sounds they produce are usually far enough after in timbre that the comb filter effect doesn't really register; we hear one group, but we still register it as a group and pick it apart that way. Even if it's one singer layering on top of themselves, the layers usually aren't close enough that the comb filter is audible. Similarly, if you have an orchestra, even though every player claims to be hitting the same notes (well, if they're supposed to), the minute differences in instrument, playing style, and even where they're sitting, cause them to yield different-enough signals that the comb filtering doesn't really appear.
So we generally consider the comb filtering effects to be really electronic sounds, unlike delay and reverb, because you don't usually hear them in acoustic life. However, if you happen to be Debbie Harris singing Heart of Glass, you sing the vocal track twice, layer the two recordings on top of each other, and end up sounding like you're singing through a phaser. The difference between the individual vocal tracks and the combined one is startling. You can really hear the amplified frequencies swinging up and down as she sings, independently of the notes she's actually hitting.

 

Encountered unexpected number theory problems when trying to optimize code.

 

Whenever someone asks why I haven't tried something before I'm going to tell them that I'm saving myself for marriage.

 

About once a year or so I learn that, despite everything, I still don't actually under the Fourier Transform.

 

I definitely have never learned the math. My take on the topic is that people who purport to explain it but don’t label their axes should give those axes back. If they don’t label entire graphs, they are banned from the frequency domain altogether. There’s very little that’s more confusing than finding what appears to be an excess time axis when you thought you were past that. Where did this time come from and what did you do with the magnitude, you monsters??

 

Yeah, spectrograms are pretty badly explained a lot of the time. Usually what they want to show is how the frequency content of a signal changes over time. So they have frequency on one axis and time on another axis, and then the magnitude of each frequency at a given moment is indicated by how dark the graph is at the corresponding point. If you're not used to this kind of picture, it's really not that informative.

As for the math, what I'm finding out (repeatedly) is that the Fourier Transform is one of those things that can be interpreted in so many ways that the natural "generalizations" of it can get arbitrarily deep. Like, sure, there's the time-frequency interpretation, but there are interpretations where even if there's an analogue of the time domain, the corresponding "frequency domain" looks nothing like a waves repeating over the corresponding time domain.

 

A coworker told me that when I have a good idea at work, I should try to have them on Fridays, so I can write them down and then feel good about myself all weekend before coming back on Monday to discover how it was actually a really dumb idea.

Anyway, I think I've gotten rid of the number theory requirement, and also got a better optimization for my code. I did so 15 minutes before leaving work, and I'm trying not to think about it until tomorrow, because it feels too easy.

 

I'm pretty sure that I've only started using the word "womble" fairly recently. I think it's a mixture of "wobble" and "ramble".

 

Metric is an objectively better system, but I don't think Vanessa Carlton would have gotten a hit single if she had talked about walking a megameter.

 

I might have to rework my idea for what the post on Vocaloid is going to be since apparently Hatsune Miku isn't in the current version. : (

 

Apparently my dreamself is convinced that it can perform retcons. I mean, I already dream in text half the time, apparently I just sometimes have it open in Notepad rather than on AO3.
I was going to say that it was almost but not quite like lucid dreaming because my dreamself only implements dumb impulse ideas, but who am I kidding, I have dumb ideas when awake too, I just don't actually have that kind of write access to reality.

 

When I was in grad school, the most interesting question I was asked was: "if you made a rap album based on your research area, what would its title be?"
I ended up with "Nothing But Lies", the joke being that the "Lie" in the name "Lie Theory" is not pronounced like the English word "lie", but more like "lee".
Anyway, that led to me naming my blogger "Mostly Lies", since I also planned to talk about quantum groups as well. Thus, in turn, this blog gets its name from the blogger title.

I just realized that I should have called it "Misrepresentation Theory", since I was doing representation theory for Lie algebras.

 

Watching a TedTalk for political reasons. I'm not even watching it, I'm just bumping the viewcount so that it can be used in legal arguments.

 

To be perfectly honest given how they operate I wouldn't vote for anyone running the Democratic Party machinery; maybe it's just all news is bad news but it seems like all they do is fuck up the getting-Democrats-elected process on basically all levels except for getting funding and maybe not outright breaking the law; notably, this list of strengths doesn't really include getting Democrats elected.

 

Hi, I’m from Chicago. I forgot how many of our last however many governors have served federal prison terms for hellacious, life ruining, democracy-destroying corruption because it’s enough that you actually can be excused for not remembering. Guess which party they were. I know governors are supposed to be state level, but this is Illinois. I’m smiling because I’m too Midwestern to laugh in bitter despair at people’s hopes that Democrats are the good guys (TM). There are no good guys.