Anatomy of a Disquiet Junto project

I participate in Marc Weidenbaum’s Disquiet Junto whenever I have the time and the brain space. Once a week, he sends out an assignment, and you have a few days to produce a new piece of music to fit. Marc asks that you discuss your process in the track descriptions on SoundCloud, and I’m always happy to oblige. But my descriptions are usually terse. This week I thought I’d dive deep and document the whole process from soup to nuts, with screencaps and everything.

Here’s this week’s assignment, which is simpler than usual:

Please answer the following question by making an original recording: “What is the room tone of the Internet?” The length of your recording should be two minutes.

Data Centre

Beyond the guidelines set by Marc’s assignments, I have some personal parameters, the main one being that the result should sound like music, by my relatively conservative definition of the word. Many Junto participants make music that is very abstract, what I would classify as sound art. That’s cool, but it isn’t where my heart is. I want to produce an approachable and fun track that any person could enjoy, starting with my wife and son. Sometimes my Junto tracks are ambient or drone-based, but usually I like to find the beats implicit in the given sounds, to use hip-hop and EDM production techniques, and to use listener-friendly repetition and structure.

Day zero

Upon reading the assignment, I have a couple of immediate thoughts. The first is the sound of a dial-up modem. I’m immediately suspicious of this idea, since it’s such an obvious cliche. My next thought is that someone must be doing sonifications of internet traffic, or converting web page source code into audio somehow. It also occurs to me that since every recording ever made is on the internet, maybe a random sampling of them or a wall of white noise would be appropriate. That idea doesn’t appeal much to me aesthetically. I’ll have to see what I can find out there.

Day one

My first move is to do a Google search for “internet sonification.” It turns up a bunch of academic papers, a browser for the blind that I can’t figure out how to even download, much less install, and many sonifications of things not related to the internet except insofar as they have web pages about them. Okay, clearly this is not the right search term.

A Google search for “sound of the internet” is more promising. There’s some intriguing music by Andrew Womack, though the relationship between his chosen sounds and their subject matter isn’t totally obvious. I also find this insane web site that randomly loads pages with autoplaying audio. The rest of the results are nostalgic recordings of modems and other old-timey computer sounds.

Next I head over to freesound.org and search for “internet.” I get more dial-up modems, of which the best two are by 0ktober and x.3d.gime. I also find this totally mesmerizing recording of a server by pm_b. There’s a guy who used a contact mic to record his hard drive spinning while he did a Google search, which conceptually is perfect, but which sadly doesn’t sound all that interesting. I also find several copies of the late Ted Stevens’ perennially hilarious “series of tubes” speech, and a lot of typing and mouse clicking sounds, all of which are intriguing, but seem too far removed. The modems and the server are the clear winners, so I download them.

The next step is to load the samples into Ableton and see what transpires.

Loading the samples into Ableton

I solo the server sound first. It has a nice intrinsic rhythm to it under the drone, and will work great as the spine of the piece. To help me nail the rhythm down, I add an 808 kick playing a four-on-the-floor pattern.

Four on the floor kick drum

Now I can more easily parse out the beat implicit in the server drone. I find that its tempo ebbs and flows quite a bit. Rather than make myself crazy trying to make sense out of the whole thing, I decide to just grab the first few bars’ worth and quantize them.

Finding the beat in the server drone

I’m not immediately sure how to combine the two different modem sounds. After listening to them a couple of times, a solution presents itself: one has a better-sounding dial tone and touch tone sequence, and the other has a better-sounding modem handshake. I slice and dice the samples accordingly.

Slicing and dicing the modem sounds

A phrase structure is emerging: two bars of dialing, two bars of modem handshake, four bars of data transmission noise. I bump the tempo up to 128 bpm so that I can fit a nice even number of phrases into the apportioned two minutes. That gives me an overall length of 64 bars. The tempo and length make for a pleasingly binary combination, very computer-appropriate.

I’m concerned that eight repetitions of an eight-bar loop is going to be too much symmetry, so I add beat repeat and tempo-synced delay to both modem tracks for greater rhythmic liveliness and unpredictability.

A structure is emerging

I always give my tracks a pop song structure, even if they’re more experimental and textural like this one. It’s an easy way to make sure that the track has a satisfying ebb and flow of energy and intensity. My personal color scheme is to use yellow for intros and outtros, blue for verses, green for choruses, and orange for breakdowns and instrumentals.

Building a pop song structure

I add a mastering suite on the master channel, a combination of compression, EQ, saturation and stereo imager. My mastering professor would be appalled, but whatever, it’s an expedient way of making everything sound better. If someone else decides to master this track properly down the road, I’ll remove all the plugins. I bounce the track and turn off the computer to let things rest overnight.

Day two

Upon sitting down for an initial listen, it’s obvious that the beat repeat is overkill. I use automation to switch it off for the first few sections. I also massage the ending a bit — I need the delay and beat repeat to come off the modem sample so that it ends cleanly on a button.

The kick drum sounds okay, but there needs to be more percussion. I remember that I have an amazing recording called Symphony for Dot Matrix Printers by The User. It’s totally technologically inappropriate, but it’s just such a cool sound, perfect computer-y percussion, and I feel like I have to at least try it. I lay it into the track and it sounds absolutely right. I put on gentle beat repeat and a resonator to disguise the origin of the sound a bit and to get it to sit better with the rest of the groove. I also split it in half so the breakdown section is still just the kick drum, that’s sounding good as is.

Finishing touches

While digging up the dot matrix sound, I discover that I also have a campy sound effect of a sixties-style computer bleeping and blooping. Maybe that’s getting too retro and ironic? Whatever, worth a listen. I also unearth a glitch recording I made named “insect computer,” a normal piece of music that got mangled by a failed file conversion in a way that turned out awesome. I lay them both in a track and audition them. Both sound great, though the sixties computer needs to be slowed down, and the insect computer needs to be pitch shifted down a minor third to fit the server drone.

I’m still worried about the camp factor in the sixties computer. Then I realize that I can get a similar effect in a much cooler way using Buchla synth samples I recorded for Morton Subotnick’s class. In particular, I’m thinking of some stuff from “Buchla E Fresh” where I had pitches and durations controlled by random voltage generator. They sound impressionistically like information being processed. I’ll have to audition those too. I don’t want to get carried away here, it’s only a two minute track, but this is all worth at least hearing.

Auditioning more sounds

I find two great samples of Buchla bleeps. I can’t decide which one I like better. I end up using one as the modulator to vocode the other.

Now it’s time to do some tidying up. I tune the kick so it plays the same pitch as the server drone. I do a rough mix — the kicks, server and modems are the foreground, and the printers, glitch and bleeps go in the background.

Mixing

I do a little more slicing and dicing to give the structure some shape, and the track is done.

Finished product

I bounce and put the track in iTunes to add metadata. Before I upload to SoundCloud, I need a title. I decide to let Ted Stevens find his way in after all and name the track “Song of the Tubes.” Drop me a line if you want to hear the finished product.

All told, this was around two and a half hours of work, not counting blogging. That’s more time than I usually put into a Junto project, but this one is getting so much documentation that I figured it was worth the extra push.