A music-first workflow consists of making the actual music itself first (the melodies, harmonies, and rhythms — anything you can describe with notation or basic MIDI events), and then later on working on the sound design.
The rough steps are:
- Compose a rough demo
- Record the fully fledged version
- Mix it down
Personally, I make a rough demo in a DAW that I'm comfortable composing in, using multiple instances of a simple plug-in. One that's complex enough that I can set an appropriate envelope for the VCA, and wire up the mod wheel to control one of several things, usually the filter's cutoff point, like I'll do with the modular synth in the final recording; but one that's simple enough that I won't be tempted to try to make any recording-worthy sounds on it. It only needs to make a rough approximation of each patch — something that's as good as it needs to be for demos, and no better.
That way, I can focus on composing the music itself: building up chord progressions, layering up more parts, maybe switching between several sections and smoothing out the seams between them. When I've finished composing this rough demo, I export it to give to my patrons, barely more than a chiptune. Then I leave it for a few days.
Once I've made a whole song this way, and I'm happy with the compositional side of it, I'll export a Standard MIDI File of all the notes and mod wheel changes, and import that into a DAW that I'm more comfortable multitrack recording in.
I use the multitracking DAW to play back the MIDI into hardware synthesisers, mostly modular, and mostly monophonic. I patch up and perform each part one at a time, until I've performed the entire song. Then I leave it another day or two.
Finally, I mix it down. I insert and freeze delays on parts that benefit from them. I adjust the level and panning of each part to get a good balance. I send some parts to the reverb bus. I try not to respect the source material, and allow myself to make radical changes in the edit. For instance, I might sidechain gate a pad to a new hi-hat part to add rhythmic interest.
At any stage, I might go back and make some amendments to the previous stage (perhaps I forgot to give my drawn-on notes different velocities, or I realise a particular point in the song is too repetitive and would benefit from adding another layer). But I never get carried away with sound design while composing, let alone start mixing so early on.
Generally, the centrepieces of a music-first studio are a multitrack recorder and a mixing desk. A DAW can cover both of these.
You can optionally have a smaller, even mobile setup for writing demos on first. Remember, they don't need to sound good.
Even if you want to avoid modern DAWs while still having quantisation and the like, you can still compose music-first with a 1980s/1990s style studio setup: try composing with a MIDI sequencer controlling a cheap multitimbral rompler for the demo, then switching out the preset sounds for custom samples later on.
If you compose your demo in a sequencer or DAW that can export the whole song's MIDI data as a Standard MIDI File that actually plays the song from beginning to end (Ableton Live currently can't, alas, but Reason can), then you can compose demos and record fully fledged songs using completely different setups. Your composing rig needs a MIDI controller keyboard (unless you draw on the notes), whereas your fully fledged studio rather perversely doesn't. So if you're out and about a lot (say, touring), then it might be useful to be able to compose on the move, even if you can only record at home.
Music-first workflow advantages
- You can tailor each sound for the one specific song it appears in
- If you record one part at a time, you can dynamically change each timbre throughout the song, adding depth and interest
- Listening to the music as a simplified mockup, you can make sure it's pulling its weight, not relying on the sound design
- You can write demos right now, even if you don't yet have good equipment to perform the music on