How do you compose?

When I compose a song, I basically use one of these approaches:

1) Substractive approach

- I make a groove which can be from 8 to 32 bars and many parts (drums, bass, pads, keys, etc.).

- Copy & paste the loop quite a few times to fill the length of the song or a section of the song.

- Start deleting clips in order to make the song evolve. This is like carving the song out of a big block of repetitive loops.

- Add a few arrangements (drum rolls, automation, transitions), and maybe even a couple more tracks with keyboard riffs or something.


2) Linear approach

- Pick a sound, arm a track, hit record, and play.

- Repeat for other tracks until the song feels complete.


3) Live approach (I haven't done this myself, but I'd like to)

- Set up a bunch of loops to play together (just like the first step of the Substractive approach)

- Set all levels to zero.

- Hit playback. Tweak the levels in order to bring parts in & out. Tweak the synths. Tweak the fx. Tweak everything. Record the result. A MIDI controller (ie.- Evolution UC-33) would be handy.


Anyone relates?

Let’s see. I pretty much make things as I go (play or mouse whatever I feel like.). I think well what would sound good here? Then I try to make what I am thinking as close as possible. Usually the drum hit's that I'm using really shape the song as they need to be complimented by the synth parts. Of course the melody must be complimented by the rhythm unless I really want to go against it. So each part writes each other. I never re-use old loops......I probably should, it would make my work flow a lot faster. Of course when I make a song with vocals it's a much more drawn out and annoying process and I don't want to get into it as I'm still figuring things out.

I am a little hesitant to answer so ambiguously, but I have to be honest, and say that I use all three approaches to varying degrees, as well as some others not mentioned. It really depends on the final objective. The best results usually arrive in my head as complete: pure inspiration, that I reverse-engineer in an attempt to recreate them in a concrete form. That's not to say that the structure doesn't change as the project develops; sometimes radically so. I leave a lot of room for tweaking and 'happy accidents', but the song itself is most often something that's screaming to escape the confines of my skull.

I suppose that if I had to generalize, I'd say that most melodic lines begin life using Method #2, although I consider that one to be a "live" approach myself. Let the 'tape' roll, and see what develops. I might then use a bit of Method #1 to weed out the obvious trash, or surgically go in and selectively edit/revise certain passages, single notes, or phrases. Rhythm section parts are often built up with Method #1, although at times I'll place a "stand-in" loop or riff to just hold the feel together. This step is essential to me, because I utilize a lot of changes and odd time signatures. Then the final tracks get swapped in later.

I also use a lot of algorithmic compositional techniques, especially if I'm emulating some natural sound occurence that serves as the foundation for the piece. That really doesn't fall neatly into any of the categories mentioned above. We should also consider an additive approach, where you start with a simple melody, and then build it up layer by layer. I'll do that with multi-tracked guitars, but instead of the usual double-tracking, I'll construct the lines from overlapping phrasing and a ton of layering, as if several dozen guitarists shared one collective mind.

As for your Method #3, I have done that, but it's the rarest of the three methods for my compositional needs. I'm fortunate to have a lot of hardware/software mixing tools at my disposal, but I'd like to share one technique that I use with Method #3 in order to add some performance interaction and a touch of randomness to it.

After you setup your project as described, use MIDI remote Control to automate the mixdown process. You have very fine control if you use automation curves, but what I'm suggesting is to mix "on-the-fly." If you use Note On control for each of your track levels, the behavior will be along the lines of a latching control. One key press turns the track full-on; the second full-off. If you choose Note On/Off, one Note press sends the level to +6 dB; the release of that key [Note Off] sends you back down to -Inf. That's more of a gating mode. Now you add a different Note for each track, and you can mix down the entire composition with a two-octave controller [< 26 tracks, anyway]. Leave one Note On location for the Master volume in your synth or the Mains, and you can bring the entire synth or mix in & out.

This technique holds some great promise for shaping your track lines, and the results can be recorded as track automation. Don't like how it turned out? Delete the new automation pattern, and mix it again, or edit the results if it isn't too far off. I like to program some particularly busy drum patterns, and then thin them out with Remote Control. It sounds great, and you can get a lot of mileage out of one stretched pattern. You've got to try this one to believe the astounding results. Remember to adjust your synth volumes accordingly, as you're now jumping from minimum to maximum level in one fell swoop. There are no 'in-between' settings.

You can also use velocity & aftertouch to impart a bit of randomness to the mix, but I'll warn you that it's difficult to control. That's is the whole point, after all. Using the sliders on your controller to send MIDI messages for mixdown is an obvious choice here, but I mention the other messages in case someone might have overlooked the possibilities.

Sheesh! I can't even answer a simple question without turning the answer into a mini-tutorial. I need to seek out some professional help for this ... Great topic, though.

Edited For Inaccuracies

kammeyjd,

I feel your pain. I compose the exact same way...slogging thru the quagmire of sounds to jog some inspiring grooves....I agree....we aren't efficient at all.
And even worse, I end up "mothering" and "nurturing" these sounds in order to develop them to such an extent that I end up using 4 or 5 FX on EACH LEAD SOUND!! That means my CPU usage is at almost 70% before my drums or backgrounds get added!!!
Thank God for Athlon 64 and all the goodies on this PC!!!


edit: Dont believe B-rock, he doesn't compose music!!! (only tutorials!)

After you setup your project as described, use MIDI remote Control to automate the mixdown process. You have very fine control if you use automation curves, but what I'm suggesting is to mix "on-the-fly." If you use Note On control for each of your track levels, the behavior will be along the lines of a latching control. One key press turns the track full-on; the second full-off. If you choose Note On/Off, one Note press sends the level to +6 dB; the release of that key [Note Off] sends you back down to -Inf. That's more of a gating mode. Now you add a different Note for each track, and you can mix down the entire composition with a two-octave controller [< 26 tracks, anyway]. Leave one Note On location for the Master volume in your synth or the Mains, and you can bring the entire synth or mix in & out.

Thanks for this. That's cool.

You know, I recently made a song, nothing but various tracks of percussion, and I assigned each track's mute button to the 8 knobs on my Oxygen8. I then recorded my knob movements, muting and unmuting the various tracks. As I was doing it, I thought, 'this is kinda stupid, I've got 8 knobs but I can only control two tracks at a time. Only two hands.'

Believe it or not, it never occured to me to assign a KEY to each track's mute button. (Is this really possible?) This way I can use all ten fingers and control more tracks at once. Duh!

Sounds like it might take some practice though.

I'm gonna try this better technique next.

ORIGINAL: Digital Aura

I'm more like Tristan I guess....

I usually start with some idea in my head of the "type of song" that I want to accomplish (ie. fast dance, chill, ambient) and work towards that.

Yes, and an eternity later...

"My First Song!"



What I do is just play. Like a kid. Play with a sound and see where it can take me. Play with some loops to see how they might fit together or progress. I look for happy accidents, and must admit, those are frequently the most productive.

how i compose...hmmm...

i usually dont have much of an idea except that i want a cool song
i load up some synth and find a preset i like and play around with it for a bit...screwing around on the p.seq because i have no keyboard. then when i make something cool, i build drums and add other instruments w/ somethimes an alternate thing for some of the other tracks...and thats how my songs fall into place.

im kinda teh "hey, that was cool! lets make a song around that." person

waiting anxiously for p5v2
syn.ops

i'd say my approach is a combination of the subtractive and linear...

ORIGINAL: Digital Aura

ORIGINAL: mike85021

Yes, and an eternity later...

"My First Song!"

...he's Baa-aack

Yes. Yes I am.

BTW - how did it work out between you and the walruss?

ORIGINAL: rabeach

i use the side variant of the reductive tripolar retrograde in combination with the selective randomized serendipitous synchronicity factor. yea what mike said.

But you expressed it so much more eloquently! Of course I use anti-bacterial low pass filters, and take medication for at least the bi-polar thingy.

Thanks for the tip, B-Rock. I'll try that soon.

ORIGINAL: syn.ops

how i compose...hmmm...

i usually dont have much of an idea except that i want a cool song
i load up some synth and find a preset i like and play around with it for a bit...screwing around on the p.seq because i have no keyboard. then when i make something cool, i build drums and add other instruments w/ somethimes an alternate thing for some of the other tracks...and thats how my songs fall into place.

im kinda teh "hey, that was cool! lets make a song around that." person
...
syn.ops

Ditto. Do you have a native instrument (i.e. - do you play keyboard, guitar etc)? A cheap keyboard would add about 100% to your musical experience, since you can record directly or step record into P-Seq. It woudl be a lot easier, too!