A vocalist sings a verse at conversational intensity, and the recording sounds warm and present. Then the chorus arrives. She pushes into her upper register, and a narrow band around 3 kHz turns piercing. Cut that frequency with a conventional equaliser and the chorus improves, but the verse loses the presence that made it intimate. The problem was never the frequency itself. It was the frequency at a certain level.

Dynamic EQ was built to close this gap. The idea has analog ancestors in de-essers and frequency-conscious compressors, but digital processing made it routine. Modern equalisers such as FabFilter Pro-Q let nearly any band become conditional. Spectral tools like oeksound's soothe extend the logic to dozens of resonances tracked at once. What was once a specialist mastering technique now sits in the default toolkit of bedroom producers, available on every channel of every session.

The shift is more than convenience. Pierre Schaeffer argued that a recorded sound should be heard as an object whose shape unfolds in time, not as a fixed spectral snapshot. Static EQ treats tone as a constant property of a source. Dynamic EQ treats it as behaviour, something that changes with pitch, intensity, and context. Understanding that difference changes how you hear problems in a mix. It also changes how precisely you can fix them without collateral damage to everything that was already working.

A conventional equaliser is a fixed filter. Set a 4 dB cut at 3 kHz and every moment of the signal receives exactly 4 dB of attenuation at that frequency, whether the passage is a whispered verse or a full-throated climax. The filter has no awareness of the audio passing through it. It shapes the spectrum the same way regardless of what the music is doing. That predictability is a virtue for broad tonal decisions, but it becomes a liability when the problem is intermittent.

Dynamic EQ attaches a detector to each band. The detector measures energy within the band's frequency range, and when that energy crosses a threshold, the filter gain begins to move. Below the threshold, the band sits at zero and the signal passes untouched. Above it, the cut deepens in proportion to how far the level overshoots, governed by a ratio or a maximum range. Attack and release times determine how quickly the filter responds and recovers.

The logic is flexible. A band can cut when energy rises above the threshold, which is the most common use, or boost when energy falls below it, lifting a weak element only when it fades. Some designs allow upward expansion, exaggerating peaks rather than restraining them. The detector can also be fed from a different filter shape or an external sidechain, so one instrument can control the tone of another. In effect, each band becomes a small compressor or expander with its own spectral address.

This differs from multiband compression in an important way. A multiband compressor splits the signal with crossover filters into adjacent regions, compresses each, and sums them back together. The crossovers themselves can colour the sound, and the bands are typically wide. Dynamic EQ uses the familiar bell, shelf, and notch shapes of a parametric equaliser, so processing can be surgically narrow while everything outside the curve remains genuinely unprocessed. There is no reconstruction of the full signal, only a localised intervention.

There are trade-offs. Minimum-phase filters shift phase as their gain changes, and fast-moving filters can introduce audible modulation, especially at low frequencies, where a single waveform cycle can outlast a short release time. Linear-phase modes avoid phase shift at the cost of latency and pre-ringing. Yet the conceptual leap outweighs these details. Frequency correction stops being a permanent statement about a sound and becomes a rule that waits quietly for the problem to appear.

Takeaway

A static EQ setting is a claim about what a sound is; a dynamic EQ setting is a claim about when a sound misbehaves. Learning to tell those two situations apart is most of the skill.

Resonances are frequencies where a system rings more readily than at neighbouring frequencies. An acoustic guitar body booms near its air resonance, a snare drum rings at the pitches of its shell and heads, a small room reinforces wavelengths that fit its dimensions, and a vocalist's formants concentrate energy in characteristic bands. These resonances are part of what makes an instrument sound like itself. They become problems only when they dominate. Removing them entirely tends to remove identity along with the fault.

The crucial detail is that most resonances are excited selectively. A room mode at roughly 110 Hz will bloom when a bassist plays an A2, whose fundamental sits right on that frequency, but stay comparatively quiet during other notes. A guitar may sound balanced across most of its range and suddenly thump on two or three pitches. The resonance is always present as a potential, yet it becomes audible only as an event. That intermittency is exactly what a static filter cannot follow.

A static notch at 110 Hz would fix the offending note but thin out every other moment in the performance, carving a permanent hole into the low end. A dynamic notch at the same frequency, with a narrow Q and a threshold set just below the peak level of the problem note, does nothing until that note arrives. When it does, the band dips, the boom is contained, and the surrounding tonal balance is preserved for the rest of the part.

Settings matter here. Narrow resonances usually need a high Q and a relatively fast attack so the filter catches the onset of the ring. Release should be long enough to avoid fluttering, particularly in low frequencies, but short enough that the band recovers before the next phrase. Soloing helps locate the problem, but final decisions belong in context, where a resonance that seemed glaring alone may be partly masked by the arrangement. Precision without context tends toward overcorrection.

Spectral processors have extended this idea to its logical extreme, tracking and attenuating many resonances simultaneously and adaptively. The results can be remarkably smooth, and that smoothness has quietly become an aesthetic in contemporary pop and electronic production. It is worth noticing the cost. Some harshness is energy, and some resonance is character. A tool capable of removing every peak raises a creative question that technical skill alone cannot answer: which irregularities are flaws, and which are the sound itself?

Takeaway

A resonance is less a fixed feature of a sound than something that happens to it under certain conditions. Correct the occurrence, and you can keep the character that produced it.

The de-esser is the oldest and most familiar dynamic EQ, even if it predates the name. Sibilant consonants concentrate energy somewhere between roughly 4 and 10 kHz, varying with the voice and microphone. A split-band de-esser reduces only that region when sibilance crosses a threshold, leaving vowels untouched. Using a full dynamic EQ for the task allows finer control: a narrow band for one particular whistling S, or a gentle shelf for airy hiss on breathy passages.

Bass consistency is another classic application. Electric basses, synth basses, and upright basses rarely produce an even level across their range. Certain notes leap forward because of instrument construction, string choice, or the room they were recorded in, while others recede. A downward band on the low fundamental region tames the loud notes. An upward band, set to boost when energy falls below threshold, can lift weaker notes, creating a steadier foundation without heavy broadband compression flattening the performance.

Dynamic EQ becomes even more powerful with external sidechaining. Feed the kick drum into a dynamic band on the bass around the kick's fundamental, and the bass yields that region only at the instant the kick lands. The same principle works higher up the spectrum. A dynamic cut in the guitars' presence range, keyed by the lead vocal, opens space for intelligibility while the singer is active and returns the guitars to full brightness in the gaps between phrases.

These techniques matter more as arrangements grow denser. Contemporary production often stacks dozens of layers, with synthesisers, samples, and processed vocals competing in the same few octaves. Static EQ can only allocate space permanently, carving each element into a fixed slot. Dynamic EQ allocates space over time, letting elements share frequency territory by taking turns. In mastering, the same thinking tames cymbals that only bite in loud choruses, or low-mid buildup that appears only when every instrument plays at once.

The broader shift echoes a long trajectory in recorded music. Early engineers fought the physical limits of tape and vinyl. Contemporary producers work in a medium with near-unlimited track counts, where the scarce resource is clarity. Dynamic EQ is a response to abundance rather than limitation. It treats the mix as a negotiation unfolding moment by moment, where the right tonal balance depends on what else is happening, and the most transparent correction is the one that occurs only when needed.

Takeaway

In a dense mix, frequency space doesn't have to be divided permanently; it can be shared over time. The question shifts from where each element belongs to when it needs room.

Dynamic EQ rests on a simple recognition: tonal problems are often temporal problems. A harsh frequency, a booming note, or a crowded midrange may exist only for a moment, triggered by a particular pitch, intensity, or combination of sounds. Treating these with permanent filters fixes the moment and damages everything else.

By joining the frequency selectivity of equalisation to the time awareness of compression, dynamic EQ lets engineers correct behaviour rather than character. That is close to what Schaeffer proposed when he asked listeners to attend to the shape of sound objects as they evolve. The spectrum is not a still image. It moves, and our tools increasingly move with it.

The next steps are already visible: adaptive processors that learn a source's typical spectrum and intervene only on deviations, and analysis that models masking between tracks. The opportunity is extraordinary precision. The responsibility is deciding which irregularities deserve to survive, because the most musical correction often knows when to do nothing.