The hook
A volunteer inherits a dull-sounding vocal channel and starts fixing it the intuitive way. Vocal's not crisp? Boost the highs. Now it's thin — boost the body back. Now it's a little honky — well, more presence will cover that. Twenty minutes later every band on the EQ is up somewhere between 3 and 8 dB, the channel is 6 dB hotter than it started, the meter's flirting with the red, and the vocal sounds... loud. Loud, and somehow still wrong.
Here's the move nobody guesses on their own: when something sounds wrong, the fix is almost never adding more of what's missing. It's removing what's in the way. The dull vocal didn't need more highs. It needed less 350.
The lesson
Think about what "dull" actually means on the map from 4.1. A vocal sounds dull when the presence region (2–6 kHz) is quiet relative to the body and low mids. There are two ways to change a relative relationship: raise the quiet part, or lower the loud part. Same balance either way. But the two moves are not equal, and the whole craft of EQ tips toward the second one.
Reason one: the problem is usually an excess. Microphones up close exaggerate lows (the proximity effect, from 2.5). Stages rumble. Small rooms pile up low mids. Almost every real-world tone problem is too much of something: mud, boom, honk, harshness. Rarely a genuine hole. Cutting removes the actual problem; boosting builds a taller mix around it and leaves the problem inside.
Reason two: boosts are loud, and loud lies. From lesson 1.3: your ears reliably prefer whichever version is louder, regardless of quality. Every boost makes the channel louder, so every boost sounds like an improvement for a few seconds. That's exactly how our volunteer walked himself into six stacked boosts. Cuts make the channel slightly quieter, so a cut has to actually fix the problem to sound better. Cutting keeps your judgment honest almost automatically.
Reason three: headroom and feedback. Every dB of boost is a dB of extra level driving the channel toward the ceiling from 2.2, and in the boosted band, a dB less margin before the room starts to ring (0.4's loop, formalized in 4.4). Cuts spend nothing. In a live room, boosts are debt.
That last phrase feels like a magic trick the first time, so here's the mechanism plainly: loud frequency regions mask quieter ones nearby. Your ear genuinely can't hear detail hiding under a pile-up (1.3). Cut the 350 Hz mud out of that dull vocal and the presence you were about to boost appears, because it was there all along, buried. You didn't add clarity. You uncovered it.
The technique: sweep, find, cut
So how do you find the exact frequency to cut? The classic hunt, on any parametric EQ:
- Name the suspect first. Use the 4.1 map — "boxy" points at 250–500 before you've touched a knob. The sweep confirms; the map aims.
- Boost a bell moderately (+6 to +9 dB, medium-narrow Q) as a temporary magnifying glass.
- Sweep it slowly through the suspect zone while the source plays. Listen for the spot where the badness blooms: where the honk honks hardest, the boom booms worst. Your ear finds it reliably; ugliness under magnification is unmistakable.
- Land there, flip the gain to a cut. Start around −3 to −6 dB, widen the Q a touch (problems are usually a region, not a single hair), and adjust by ear.
- Toggle the band in and out. In: problem gone? Out: problem back? If bypassing it barely changes anything, you don't need the cut. Delete it and keep the channel simple.
Go deeperoptional
The masking mechanics, one level down. On the basilar membrane, a strong tone excites a region, not a point, and the excitation pattern spreads asymmetrically: it reaches further toward higher frequencies than lower ones ("upward spread of masking," 9.3 territory). Strong low-mid energy therefore raises the audibility threshold for content above it. The 350 Hz pile-up was literally raising the level that 3 kHz consonants needed to reach before your ear registered them. This is why cutting lows and low-mids so often reads as "adding" highs, and why the effect is bigger than symmetric intuition suggests: you're removing a masker that punched above its band.
Why boosts sound worse than cuts of equal size. Several effects stack. Perceptually, the ear grew up tolerating notches (every room and every head-related transfer function is full of dips) but flags narrow peaks as unnatural, because in nature a resonance usually signals something wrong; a ±6 dB bell offends more as a boost than as a cut at the same Q. Technically, a minimum-phase boost also produces more audible ringing at its center frequency: energy at the peak decays more slowly, which reads as a coloration or "whistle" at high Q (the 4.1 phase discussion, cashed out). And operationally, a boost raises the channel's peak level: a 6 dB boost can genuinely halve your remaining headroom if program energy sits in that band.
The equal-loudness trap, quantified. Loudness comparisons need matching within about 0.5–1 dB to be fair; differences beyond that reliably bias preference toward the louder version regardless of content (the research behind every "louder sounds better" A/B you've ever fallen for, including in gear stores). A 6 dB broadband boost is roughly a 40% jump in perceived loudness by the Stevens' law approximation from 1.2's deep end, far past the bias threshold. Hence the discipline: every EQ judgment is made at matched loudness or it isn't a judgment, it's a preference for volume wearing a disguise.
When boosting is right. Subtractive-first is a bias, not a religion. Legitimate boosts: broad, gentle shelves as overall tone tilts (+2 of high shelf as "air" on a dark vocal mic); presence lifts on a genuinely dark source when the excess-removal pass is already done; creative decisions in studio contexts where headroom and feedback aren't at stake. The professional shape of a channel's EQ tends toward narrow cuts, wide gentle boosts: surgical downward, musical upward. If you find a narrow, deep boost on your EQ, you're usually holding the problem by the wrong end.
Remember
- Most tone problems are an excess, not a hole. Cut the excess; don't build around it.
- What was masked comes forward on its own: cutting mud "adds" clarity you never boosted.
- The hunt: name the suspect from the map, sweep with a temporary boost, land on the bloom, flip to a modest wider cut, toggle to verify.
- Level-match every judgment. Louder always sounds better; don't let volume grade your EQ moves.
- Boosts cost headroom and feedback margin; cuts are free. In a live room, boosts are debt.
Your turn
- Run one full sweep-find-cut hunt on the pastor's mic, from named suspect to verified cut.
- Practice the level-match habit: after your cut, nudge the fader up to match and only then judge.
- Audit your board for inherited EQ: find one narrow boost someone left behind, and test whether a cut somewhere else does its job better.
- Once this week, when a channel sounds like it's missing something, force yourself to find what's in the way first. Only boost if the removal pass leaves a real hole.