Loudness matching

Louder always sounds better#

Human ears prefer louder. So if one version is louder but worse and the other is better but quieter, most people will pick the louder one.

The classic case is the before-and-after demo where the "after" is a bit louder. You think, wow, that processing is doing something magical. It isn't. It's just louder. The same applies whether you're comparing plugins, masters or DACs. So you absolutely must match before-and-after comparisons in loudness, to take loudness out as a variable and hear what the processing is actually doing.

Match by loudness, not by peak#

Peak normalising two clips puts their highest sample at the same level. It doesn't make them equally loud. A clip with a smaller crest factor will sound louder at the same peak, and then you're back to comparing loudness.

So I match by perceived loudness: LUFS normalising both, or matching carefully by ear. Both are fine. Once they play at the same loudness you can talk about the different crest factors as a separate matter. When I clipped a snare to get it louder, it sounded bigger and fuller than the original until I loudness normalised both. Then the original had the huge transient and the clipped one sounded weak and feeble. When I tested mixing with clippers, I loudness normalised every version for a fair comparison, and the clipped mix still sounded worse. More on that in clipping.

Experience does not get you out of it#

Some experienced engineers say gain matching is for beginners: once you know what a plugin does, you don't need it. I disagree. I dial a compressor in, I think it's really adding something, then I gain match it and A/B it, and quite often it isn't working. It sounds better without it, or at best it's unnecessary, so I delete it and the result is more transparent.

With dynamics, compression and limiting, I really feel you should do matched A/Bs. With EQ, experience does count for more, because you learn what a big bass boost will do to the clarity of a mix. I still often gain match big EQ moves as a sanity check. Knowing about a bias doesn't make you immune to it.

Auto makeup gain cheats in favour of the compressor#

Auto makeup gain almost always overcompensates. It makes the compressed signal louder than the original, so when you toggle the compressor on, there's a bump in level and you think it sounds great. I'd aim to set the output gain manually so bypass is at the same level.

No compression is better than compression you think you need but don't. My free compression course goes through this.

Boosts are louder, which is why boost-heavy EQ fools you#

Every EQ boost is also a loudness increase. The ear is tricked into thinking it sounds better. That's why a boost-only approach, especially in mastering, is in my opinion a red flag, and why I work with negative EQ. At the very least, I'd balance boosts with cuts and gain match the result.