I have ~3,000 voice notes on my phone. About 1% become finished songs. The rest are the soil. The capture pipeline I run — from phone memo to released track — has done more for my output than any DAW upgrade. Here it is, in detail.

Stage 1 — capture

Anywhere, anytime, no judgment. Hummed melody, two-line lyric, beat pattern tapped on a steering wheel. The rule: capture in under 10 seconds or the idea is gone. Open Voice Memos, hit record, sing or hum, save. Don’t title it; that’s later.

Stage 2 — weekly triage

Every Sunday evening, 30 minutes, I review the week’s voice notes. Three buckets:

  • Spark — there’s something here, worth developing
  • Compost — not now, but worth keeping
  • Delete — not the one

Sparks get titled and filed. Compost stays in the general voice memo pool with no title. Deletes go to trash. The triage is brutal; it has to be, or the library becomes unusable.

Stage 3 — sketching

The sparks that survive triage get a 30-minute Logic Pro session each. Not finished — just enough to capture the idea in some structured form. Voice over guitar, voice over piano, voice over a basic loop. The point is to bridge from “phone memo” to “DAW project,” because that bridge is where most ideas die.

Stage 4 — sit with them

Sketches go into a folder and I leave them alone. Some songs need to ferment for months before I know if they’re real. “X” sat for nearly four months before I picked it back up. The good ones don’t die in the folder; they get louder when you ignore them.

Stage 5 — develop or kill

Once a quarter I open the sketches folder and ask: which of these is the next single? The ones that survive get the full session, the studio time, the production. The ones that don’t either get archived or get cannibalised — the bridge from one becomes the chorus of another.

The lesson hidden in the math

3,000 voice notes → 30 sketches → 5 finished tracks per year. The funnel is brutal but the inputs are cheap. The bottleneck isn’t ideas; it’s willingness to capture without judgment. Most songwriters edit at the input layer. That’s where the catalogue dies.

Tools I use

  • Apple Voice Memos for capture
  • Logic Pro for sketching
  • A simple Notion table for the sparks
  • External SSD for archives

That’s it. No special software. The pipeline is the discipline, not the tool.

The principle

Capture is cheap. Curation is the work. The artists with the best catalogues aren’t the ones with the best ideas; they’re the ones with the best capture-to-curation ratio.

Built your own capture system? Trade notes with me.