Loudness

The listener owns the volume knob#

Once music is released, the playback level is unenforceable. Spotify does its own gain compensation, and even if you switch that off you can still tell Alexa to turn it up. On a turntable the difference between singles and LPs is so big that you are always adjusting the gain. On an iPhone the volume buttons are right there. It is a bizarre thing to assume the listener is stuck at a fixed volume.

Louder and quieter are relative terms. Louder compared to what? If the reference point is the listener's volume knob, which you have no knowledge of or control over, you are not entitled to use the words at all. So a "loud mix" means nothing.

Don't focus on loud, focus on good#

Music needs its transients to sound good. People who reach for clippers to get loud are often focused on making their music loud, not necessarily good, and the irony is that once you cut the transients off, the music sounds mushy, lifeless and lethargic. If you compare the clipped version with the original at the same loudness, which is the only fair way to compare anything (loudness matching), the clipped one is quieter and more feeble: it does not even make things louder.

Most of the time I'm more moderate in my loudness and I don't try to smash everything to infinity. Many people just want it louder and don't care about the distortion that comes with it.

LUFS is a helpful number, not a meaningful target#

If someone asks me to master an album to minus 14 LUFS, I'm probably just going to email back: you want your album to sound uneven? If you measure the LUFS of a song with a quiet intro, then cut the intro off and measure again, the value changes. The level I would set the chorus to for my listening pleasure does not. So normalising every track on an album to one number can give you inconsistent perceived levels.

LUFS was designed to approximate perceived loudness, and it gets closer than RMS because bass does not fool it so much. But designed to get closer to perception does not mean it is perception. Treating it as some infallible number we must abide by is a mistake.

Fixed reference level#

I set my monitoring at the exact same perfectly recallable level when I'm mastering. Obviously I can turn it down or up if I so choose, but when I'm actually working, it is always at the exact same level which I always return to. I've been shouting from the rooftops about recallable working monitoring gain for years. Every time you adjust the monitor gain, you mess with your perception of the tonality. Why that matters is under monitoring level.

The loudness war is older than the CD#

Arguably it goes back to the 1960s. Bob Olhsson, a mastering engineer who worked at Motown told me when I interviewed him, that even then A&R people were drawn to whatever jumped out merely because it was louder. Vinyl kept it in check, because cut level is set by groove space, not by limiting (see vinyl). Digital accelerated it through the '80s and '90s, and now streaming normalisation has calmed it down slightly but it still persists.

A loud master VS a dynamic one#

Digital masters often get pushed loud because clients complain that more dynamic masters sound weak and feeble next to everything else. My suggestion is to release the slammed version to streaming for people who put music on in the background whilst doing the hoovering, then put high-resolution masters with no limiting, or the limiting turned right down, on Bandcamp, called HR or some similarly appropriate label to differentiate it from the slammed version. I think listeners who want dynamics would appreciate the mere abililty to purchase it, rather than being stuck with a master they don't want.