Skip to content

Loud masters can shrink after matching

Normalization plays a crushed master and a dynamic one at the same loudness. Then the only difference left is how far the drums rise above the mix.

Updated · 5 min read Lesson 34 of 44 in Mixing & Mastering

In short

  1. Normalization turns each master down by its own amount until both play at the same loudness, so the level you won with the limiter is cancelled.
  2. A plain gain change keeps the distance from peak to loudness, so the dynamic master keeps its taller drum hits at playback.
  3. Compare limiter settings at matched loudness, because the louder version wins every unmatched comparison.
In this lesson (8)

The loudest track in the playlist that wasn't

You push the master until it reads -6 . In the session it hits hard. On a streaming service, next to a song mastered around -11 LUFS, yours sounds smaller: the kick has less weight, the snare less crack, and the other song seems to jump out of the speakers.

Nothing went wrong in the upload. Both songs were turned down to the same loudness, and at the same loudness the one with more room above its average sounds bigger.

Why it matters

Normalization cancels the only advantage

On Spotify's default setting, both masters are turned down until they measure -14 LUFS: yours by 8 dB, the other by 3 dB. The level race you won in the session is cancelled. What remains is what you paid for it: flatter and less distance between the hits and the space around them.

Apple says the same about its Sound Check: songs mastered loud are played back at a lower volume, and Apple warns that this can make them sound weaker (Apple, 2021).

Figure 1. Two masters at the same average power, which is roughly what normalization matches. The loud one was pushed about 9.5 dB into a hard clip and turned down. Its peaks now sit about 6 dB below the dynamic master's: a crest factor near 4 dB against 10 dB.

Listen

Hear what normalization does to a loud master

A synth loop or a real mix, as a dynamic master and a loud, clipped one. Turn on streaming-style normalization and compare them at the same loudness.

Loading demo…
100%

Start with your speakers or headphones low.

The science

The distance from peak to average

The number that matters here is how far the peaks rise above the average. For a waveform it is the :

C=20log⁡10(xpeakxRMS)C = 20 \log_{10}\left( \frac{x_{\text{peak}}}{x_{\text{RMS}}} \right)

A pure sine has a crest factor of 3 dB, because its peak is 2\sqrt{2} times its RMS level. A drum recording has far more. A or clipper lowers the peaks so you can raise everything else, and the crest factor drops. For a whole master the same idea is measured as the : minus integrated loudness.

Take two masters that both peak at -1 . Master A measures -6 LUFS, so its peaks sit 5 dB above its loudness. Master B measures -11 LUFS, so its peaks sit 10 dB above. moves every sample of a track by one fixed gain, which keeps those distances. At -14 LUFS, A's peaks reach -9 dBTP and B's reach -4 dBTP. B's loudest hits rise 5 dB further above the same average than A's, and that distance is much of what you hear as size and punch.

Figure 2. Master A at -6 LUFS and master B at -11 LUFS, both peaking at -1 dBTP, after normalization to -14 LUFS. A plain gain change keeps each distance from peak to loudness, so B's peaks rise 5 dB higher above the same average.

Try it

Match first, then judge

0 of 6 steps done

At matched loudness the dynamic master often sounds bigger: the drums stand further out and the chorus lifts more. If the loud one still wins, its density is doing musical work and you can keep some of it.

Common mistake

Comparing at different levels

Judging limiter settings without matching levels is the main trap. A version that is even 1 dB louder often sounds better in a quick comparison when nothing else has changed, so heavier limiting keeps winning in the session and losing on the platform.

The other mistake is blaming normalization. It adds nothing and removes nothing except level. If a master sounds small after normalization, it sounded just as small in the session at the same level; you were simply listening louder.

Takeaway

Spend the distance carefully

Compare limiter settings at matched loudness every time. Push further only while the song gains density you want, and stop when the drums start to shrink. On a normalized service, the distance between your peaks and your average is what survives. The rules behind the turn-down are in the lesson on streaming loudness myths.

3 questions

Check yourself

  1. Masters at -6 and -11 LUFS are normalized to -14 LUFS. How far is each turned down?
  2. Both masters peak at -1 dBTP. After normalization to -14 LUFS, where do the loud master's peaks sit?
  3. Why does the dynamic master often sound bigger once both are normalized?

Sources

Show all 3

Next lesson · 35 of 44

Impact is not the same as levelA chorus can be loud and still not land. Impact comes from distance: hits above their body and sections above each other, which limiting closes.

Previous: Streaming loudness myths waste masters

Portrait of Virzy Guns

Written by Virzy Guns

Songwriter and producer, founder of Virzy Guns Production. Now building HealingWave.