Parametric EQ

Every broadcast voice you've ever admired has been shaped: weight added here, mud carved out there, a little shine on top. The tool that made that shaping precise — the parametric equalizer — turned tone from a fixed bass-and-treble knob into an instrument you can aim. Here's where it came from, how its three parameters work, and how to point it at a human voice.

Before parameters: tone knobs and graphic EQs

Equalization is older than recording studios — the word comes from telephony, where engineers added circuits to long lines to "equalize" the frequencies the cable attenuated. In audio equipment, tone control arrived first as the familiar bass and treble knobs; Peter Baxandall's elegant 1952 feedback circuit became so standard that his name is still shorthand for that style of shelving control. Studios got more capable units — program equalizers like the Pultec EQP-1A of the 1950s offered a handful of switch-selected frequencies with continuously variable boost and cut — and live venues got the graphic equalizer, a bank of fixed bands with sliders whose positions draw a "graph" of the response curve.

All of these shared one limitation: the frequencies were chosen by the circuit designer, not by you. If the resonance you needed to tame sat between two bands of a graphic EQ, you had to cut both neighbors and take the collateral damage. Tone control was a set of preset holes; the signal's actual problems rarely lined up with them.

Massenburg's idea

The fix was articulated by George Massenburg, a young recording engineer, in a paper presented at the Audio Engineering Society convention in 1972. The paper's title coined the term: parametric equalization. The idea was to make every defining parameter of an EQ band continuously and independently adjustable — the center frequency (where it acts), the gain (how much boost or cut), and the bandwidth, or Q (how wide the action spreads). Instead of choosing among a designer's preset frequencies, the engineer could sweep a band across the spectrum, find the exact spot by ear, and then decide how hard and how wide to act on it.

It's hard to overstate how completely the idea won. Massenburg went on to build revered hardware under his own GML name, but the architecture escaped any single product: the channel EQ on virtually every mixing console built since, and every EQ plugin in every digital audio workstation, is a parametric equalizer. When software shows you draggable dots on a frequency-response curve, you're looking at Massenburg's three parameters with a modern interface.

The anatomy of a band

Parametric bands come in two shapes. A bell (or peaking) band boosts or cuts around its center frequency and tapers back to unity on both sides — the width of that bell is the Q. Q is defined as center frequency divided by bandwidth, so higher Q means narrower: a Q of 1 is a broad, musical stroke around an octave and a half wide, while a Q of 10 is a sliver suited to notching out a single resonance — taken to the extreme, that's how a hum filter removes grid tones without touching the voice. A shelf band behaves differently: above (high shelf) or below (low shelf) its corner frequency, the response flattens out at a constant gain, tilting an entire end of the spectrum up or down at once. The convention in voice work is shelves for the extremes — the lows and the "air" — and bells for surgical work in the middle.

Gain, the third parameter, is deceptively simple: a few dB is usually plenty. The ear judges tonal balance on ratios, and a 6 dB regional change is dramatic. Large boosts also raise everything living in that band — noise, room reflections, bleed — and eat into headroom, which is why EQ decisions are tangled up with gain staging.

A voice, mapped by frequency

Aiming an equalizer at speech starts with knowing where things live. The spoken voice lays out remarkably consistently:

• 100–250 Hz — the fundamentals and chest resonance. This is weight and authority; too little sounds thin and telephone-like, too much sounds boomy.
• 250–500 Hz — the mud zone. Proximity effect, desk reflections and small-room buildup pile up here, and excess reads as "boxy" or congested.
• 2–5 kHz — presence. Consonant energy and articulation live here, in the region where human hearing is most sensitive; this band is intelligibility itself.
• 5–8 kHz — sibilance. The s, t and ch sounds; enough for crispness, too much for comfort.
• 10 kHz and up — air. Openness and sheen; the difference between a voice that sounds recorded and one that sounds present.

Every practical voice EQ move is a sentence written in this vocabulary: more weight, less box, more presence, a little air. The map also explains why the same setting can't serve every voice — a deep-voiced speaker on a close condenser mic accumulates mud that a lighter voice on a distant laptop mic never has. A spectrum analyzer shows you the same map drawn live by your own voice, and it's worth watching one while you speak before deciding what to change.

Cut first: the subtractive philosophy

Experienced engineers reach for cuts before boosts, and the reasons are practical rather than aesthetic. Most tonal problems are excesses — a room mode, proximity boom, mud — and removing the excess is cleaner than boosting everything around it to compensate. Cuts don't amplify noise; boosts do. Cuts free headroom; boosts consume it. The working heuristic is: cut narrow, boost wide — surgical Q for removing problems, gentle broad strokes for flattering the result. Order matters too: EQ belongs before compression in the chain, so the compressor responds to the corrected tone rather than pumping on the boominess you were about to remove.

In DeskBroadcast

DeskBroadcast's Voice EQ is a parametric design with the frequency and Q decisions already made — four bands fixed at the landmarks of the voice map, each with only a gain control (±12 dB), so the one parameter left is the one that's hard to get wrong. Depth is a low shelf at 120 Hz for chest and weight; Body is a bell at 250 Hz with a Q of 1, sitting on the mud zone; Presence is a bell at 3.5 kHz for articulation; Air is a high shelf at 10 kHz. The built-in voice presets read like the philosophy above in miniature: Studio Booth (+5 depth, +2 body, +1 presence, −1 air, compressor on) builds a close, warm, radio-booth weight, while Podcast (+2 depth, −3 body, +3 presence, +2 air, compressor on) makes the classic subtractive move — carving mud at 250 Hz — and spends the savings on clarity and air. The EQ stage also carries the clean digital makeup gain set by mic calibration, so level correction and tone shaping happen in one precise, transparent place.