Step 6 of 13 · Polish

Mastering

Mastering is the final polish: loudness, tonal balance and consistency so your track holds up next to commercial releases on streaming. This is the biggest AI-vs-human overlap point — AI mastering (LANDR, eMastered, iZotope Ozone's assistant, CloudBounce) is fast, cheap and genuinely good for most releases; a mastering engineer still wins for flagship singles. Target the streaming standard (around -14 LUFS integrated) rather than chasing maximum loudness.

The steps

  1. 1Master from a headroom-leaving mix (peaks ~-6 dBFS, no mix-bus limiter slamming).
  2. 2Use an AI master or engineer to set loudness, EQ and stereo balance.
  3. 3Target ~-14 LUFS integrated for streaming; check on phone + earbuds + speakers.
  4. 4A/B against a reference release at matched loudness.
  5. 5Export 24-bit WAV for distribution (distributors transcode to each platform).
🤖 AI path

AI mastering is the default for AI releases — fast and inexpensive; most listeners won't tell on a good master.

🎤 Indie path

An engineer adds taste and translation for important singles; many indies use AI for album cuts and a human for the focus track.

🔗 Where they meet

Same target either way: streaming-loudness, translation across devices, 24-bit WAV out.

The verdict

AI mastering is the right default for AI releases — cheap, fast and good enough for streaming; save a human engineer for the single you're betting on.

FAQ

Is AI mastering good enough to release?

For most tracks, yes. Tools like LANDR, eMastered and Ozone produce release-ready masters cheaply. For a flagship single, an experienced engineer can still add an edge.

How loud should my master be?

Aim for roughly -14 LUFS integrated — the rough streaming reference — and prioritise translation across devices over sheer loudness, since platforms normalise volume.