The Myth of Subtractive Synthesis

[errantember]

Inspired by the recent InSoc and Cruxshadows shows as well as by Generation Ecstasy, I fired up Ableton Live (v. 5.x) tonight and started disassembling InSoc's hit single "Running." The goal was to break the song down into it's individual elements, such as the bass line, the high-hat, the vocals, etc, so that they could then be re-combined in interesting ways. One major technique I was hoping to use to better advantage is subtractive synthesis. This is where you take a noise that's easy to isolate (like the beginning bass line) and then subtract it from the rest of the song. The result? The rest of the song, without the bass line!

Simple.


By doing this repeatedly, one can strip the song apart piece-by-piece of anything that repeats, and end up with just the vocals at the end.

At least, in theory. :)

I couldn't figure out how to subtractive synthesis in Ableton, though it clearly must be possible, so I tried in Audacity (an free audio editing package) instead. Sure enough, if I take a bass line, copy it, invert it, and then re-mix it with itself, I get the expect result -- silence. However, it turns out that even with music as clearly machine-generated as Insoc's early work, not all bass-beats are created equal. Though they might sound the same to the ear, tiny changes have been made to the beats to make subsequent versions *not* *quite* the same, and even the tiniest amount of deviation makes the effect much weaker.

It's very easy to picture the sneaky Kurt Larson introducing such things just to piss off re-mixers. Although, really, it's likely the variation is mainly there to give the music a more live feel.

So much for *that* idea.

The next thing I started to try before running out of time is using an equalizer to cut out the bass, which is almost, but not entirely, low-frequency. This worked sort-of well, and I'll have to play with it more later.

Comments

[dj-warhammer.livejournal.com]

The last thing an electronic musician would want to do is piss off remix artists. Beat variations exist due to machine sensitivity and error, to be sure, but mainly because the artist and producer have tweaked it to make it sound aurally perfect. The beat wouldn't look perfectly even on a visual curve but it sounds just right.

[errantember.livejournal.com]

Ok, cool. I'm sure I'll have more questions about a lot of this as I move forward. Your DJing and re-mixing expertise should be a big help.

What other techniques do you use to isolate a certain part of a song, especially when it's mixed in with something else you don't want?

[dj-warhammer.livejournal.com]

Jacob would be a lot better to talk to than I. He's vastly more experienced at doing such things. I can send you his email if you don't have it already.

I'm much more focused on live performances, mixing and overlaying tracks using my board. It's possible to achieve some of the effects you're talking about just by using the EQ. Cut the mids and treble and you're going to get an emphatic bass line with very little else.

I do prepare songs in advance for "live" play. Audacity is useful for removing extended intros, outros, and jarring wonky bits that musicians stick on there randomly to amuse themselves. Other than that, what you hear is me mixing on the fly.