Mastering EQ starts with the big picture

The notch that ate the bass line

The master sounds a little dark and heavy. You open an analyzer and see a tall spike at 55 Hz and a few bumps in the low mids, so you notch each one. The analyzer looks smoother. The master sounds thinner, the bass line now jumps out on some notes and hides on others, and it still sounds dark.

The spike at 55 Hz was the bass player's A, the root of the song. The darkness was the slope of the whole spectrum, which no single notch can change.

Why it matters

Broad problems and narrow problems need different tools

On a master, the first question is which kind of problem you have. A broad one (too dark, too bright, too heavy) is heard all through the song and in every instrument. A narrow one, such as a ringing snare or a harsh resonance in a guitar, sits at one frequency whatever the music is doing. A broad problem wants a broad, gentle move. A narrow problem may justify a narrow cut, but only once you know it is a defect and not part of the song.

An analyzer cannot tell you which is which. It shows where the energy is, and in music much of that energy is the notes being played.

The science

Tilt, bandwidth and what a peak means

Released music shares a broad long-term shape. Pestana and colleagues (2013) analysed the long-term spectra of a large set of popular commercial recordings from 1950 to 2010 and found a consistent leaning toward a target curve shaped by industry practice. A master that sounds dark or bright usually differs from that shape in its slope, so the usual fix is a tilt: lower one end and raise the other around a pivot.

Bandwidth decides how much of the music a move touches. For a bell, Q and bandwidth $BW$ in octaves are related by (Toy, 2021):

A Q of 0.7 spans about 1.9 octaves, the kind of small, broad move the lesson on mastering and translation shows. A Q of 8 spans about 0.18 octave, roughly a whole tone. That is narrow enough to catch a single note, which is the problem: a 4 dB cut at 55 Hz turns down every A the bass plays and leaves the D at 73 Hz almost untouched. You have rebalanced the bass line from the master.

The test for a resonance is movement. A peak that comes from notes moves when the chords change. A peak that sits at the same frequency under every chord, and that you can hear as a ring or an edge, is a candidate for a narrow cut.

Level matters as much as shape. Loudness meters following ITU-R BS.1770 weight the upper frequencies up by about 4 dB before measuring, and a brighter master also tends to sound more exciting on first listen. A high-shelf boost therefore raises both the reading and the impression. Match short-term loudness after every EQ move before you judge it.

The demo plays chords through one filter with the spectrum drawn live. Watch the peaks the chord notes make, then sweep the cutoff: the whole sound gets darker or brighter together. Raise the resonance and one narrow region starts to stick out.

The reference you compare against sets the target slope, and that target has moved over time. Hove, Vuust and Stupacher (2019) measured songs from the Billboard Hot 100 between 1955 and 2016. Loudness rose over the period, and when they controlled for overall level, only the lowest frequency bands showed an increase. Two records with the same genre tag but thirty years apart can disagree about how much low end is right, so choose references from the era and substyle of the record you are making.