traditional approaches to resonance

the tool you reach for first

you have a vocal track that sounds harsh. your instinct is to reach for an EQ, sweep through the midrange until you find the offending frequency, and pull it down. this is the right instinct. EQ is the oldest, most understood, and most widely available tool for shaping frequency content. for many resonance problems, it is all you need.

but not all resonance problems are the same. some are constant. some come and go. some sit at a fixed frequency. some shift with every note. traditional approaches to resonance control, including static EQ, dynamic EQ, and multiband compression, each handle a specific subset of these problems well. understanding which tool matches which problem saves you from over-processing, under-processing, or reaching for the wrong tool entirely.

static EQ

a static EQ cut is a permanent change to the frequency response. you choose a frequency, a bandwidth (Q), and a gain reduction. the cut is always active, regardless of what the signal is doing.

this works when the resonance is consistent: a room mode that colors every recording made in that space, a microphone with a fixed presence peak at 4 kHz, a synth patch with a static resonance in the filter. the problem is always there, so the fix can always be there.

practical approach:

• find the frequency: boost a narrow bell (Q 6-10) by 6-10 dB and sweep slowly through the 1-6 kHz range. when the harshness gets dramatically worse, you have found it
• cut conservatively: pull down 2-4 dB with a moderate Q (3-6). listen for the harshness to reduce without the source losing presence
• check in context: solo the track, apply the cut, then unsolo and listen in the full mix. a cut that sounds right in solo can sound too aggressive in context, or vice versa

the limitation of static EQ is permanent collateral damage. the 3-4 kHz range is also where vocal consonants, guitar pick attack, and snare crack live. a static cut removes the resonance but also removes energy that was not a problem. on quiet, well-behaved phrases, the cut makes the source sound duller than it should.

dynamic EQ

dynamic EQ solves the biggest problem with static EQ: it only acts when the resonance is actually present.

a dynamic EQ band has a threshold. below the threshold, the band is inactive and the frequency response is flat. when the signal energy at that frequency exceeds the threshold, the band activates and applies the specified cut. when the energy drops back below threshold, the band releases and the response returns to flat.

this means quiet, well-behaved phrases get no processing at all. the vocal retains its full presence and clarity. only when the signal pushes into the harsh range does the dynamic EQ step in, applying just enough reduction to tame the peak, then stepping back out.

practical settings for vocal resonance:

• frequency: same sweep technique as static EQ, targeting the 2-5 kHz range
• Q: moderate (3-5). too narrow and pitch shifts cause the resonance to dodge the band
• threshold: set so the band engages only on the harsh moments. listen to a loud phrase and adjust until the gain reduction meter shows 2-4 dB on the worst peaks
• ratio: 2:1 to 4:1. ratios above 6:1 approach limiting and can sound unnatural on tonal material
• attack: 1-5 ms (fast enough to catch the onset of the harsh moment)
• release: 20-50 ms (smooth enough to avoid pumping, fast enough to recover between syllables)

multiband compression

multiband compression takes a different approach. instead of targeting a specific frequency with a gain change, it splits the signal into frequency bands and applies compression independently to each.

a typical multiband compressor splits the spectrum into 3-6 bands. each band has its own threshold, ratio, attack, and release. when the signal in a band exceeds its threshold, the compressor reduces the level of everything in that band.

for resonance control, you would set the band boundaries to isolate the problem range (say, 2-5 kHz), set a moderate threshold and ratio, and let the compressor duck the entire band when the resonance pushes above the threshold.

the problem: multiband compression is indiscriminate within each band. it does not distinguish between the resonant peak and the useful content sitting at the same frequency. when a 3 kHz resonance triggers the compressor, everything between 2 and 5 kHz gets reduced: the resonance, the vocal presence, the guitar attack, the snare crack. the compressor does not know the difference.

this makes multiband compression a blunt tool for resonance. it is effective for broad tonal balancing (taming an overly bright mix, controlling low-end buildup) but poorly suited for surgical resonance removal.

when each tool works

the three tools solve different problems. choosing the wrong one is not just inefficient; it actively degrades the signal.

static EQ works when:

• the resonance is consistent and always present
• the problem sits at a fixed frequency that does not shift
• you can afford to permanently reduce energy at that frequency
• one or two narrow cuts are enough to solve the problem

dynamic EQ works when:

• the resonance appears intermittently (louder phrases, certain notes)
• you can identify 1-4 specific problem frequencies
• preserving the natural dynamics of the source matters
• the problem frequencies are relatively stable (they do not shift by more than a semitone or two)

multiband compression works when:

• the problem is broad tonal imbalance, not a specific resonance
• the entire frequency range needs dynamic control (e.g., an overly bright mix bus)
• you are already using multiband for other purposes and can address a mild resonance as part of the same processing

where traditional tools fall short

there are three scenarios where static EQ, dynamic EQ, and multiband compression all struggle.

shifting resonances

a singer moving through different vowels shifts formant frequencies across hundreds of hertz. an “ah” vowel might resonate at 2.8 kHz. an “ee” vowel shifts that resonance to 3.5 kHz. a dynamic EQ band parked at 3 kHz catches the “ah” but misses the “ee” entirely. you can add a second band at 3.5 kHz, but then the singer moves to “oh” at 2.2 kHz. you are chasing the problem with fixed-position bands while the target keeps moving.

too many simultaneous resonances

a bright acoustic guitar has resonances at the body frequency (around 200-300 Hz), the bridge resonance (1-2 kHz), and several string overtone buildups (3-6 kHz). that is 5-7 simultaneous resonant peaks. a 4-band dynamic EQ cannot cover all of them. an 8-band parametric can, but setting 8 bands by ear and adjusting thresholds for each is tedious and fragile. change the guitar, and the resonance positions shift.

resonances that interact

fixing one resonance can reveal another. cutting a 3 kHz peak makes a 4.5 kHz peak that was previously masked become audible. you cut 4.5 kHz, and now a 2 kHz buildup is revealed. this whack-a-mole pattern is a sign that the problem is systemic, not localized.

signal chain order

where you place resonance control in the signal chain matters.

before compression: reducing resonant peaks first gives the compressor a smoother signal to work with. this prevents the resonance from triggering the compressor’s detector, which would cause the entire signal to duck every time the resonance spikes. this is usually the right order for individual tracks.

after compression: the compressor may have introduced or emphasized resonances (compressors can boost the relative level of resonant frequencies during release). if harshness appears after compression, place the resonance control after it.

on the mix bus: apply resonance control gently and only to address problems that survive individual track treatment. mix bus processing affects every element in the mix. a 3 dB cut on the bus reduces not just the harsh vocal but also the snare presence and the guitar attack.

the step beyond traditional tools

static EQ, dynamic EQ, and multiband compression cover the majority of resonance problems you will encounter. for consistent, localized resonances, they are faster and more predictable than spectral processing.

but when resonances are numerous, shifting, or systemic, traditional tools hit a ceiling. you run out of bands. you spend more time adjusting parameters than mixing. the trade-offs stack up.

this is where spectral processing enters the picture: tools that analyze hundreds or thousands of frequency points simultaneously and apply gain changes to each one independently. they do not need you to find the problem frequencies. they find them automatically, every few milliseconds, and adjust in real time.

the right approach depends on the problem. most of the time, the simplest tool wins.

frequently asked questions

frequently asked questions

should you use static EQ or dynamic EQ for resonance?

it depends on whether the resonance is constant or intermittent. if the problem frequency is always present at the same level (a room mode, a microphone presence peak), static EQ is simpler and more predictable. if the resonance only appears on louder phrases or certain notes, dynamic EQ is the better choice because it only cuts when the problem actually occurs, preserving clarity the rest of the time.

can multiband compression fix resonance problems?

multiband compression can reduce resonance by compressing the band that contains the problem frequency. however, it affects everything in that band, not just the resonant peak. if the resonance sits in the 2-4 kHz range, the compressor also reduces vocal presence, guitar attack, and cymbal detail in the same range. it is a blunt instrument for a surgical problem.

how many EQ bands do you need for resonance control?

most parametric EQs give you 4-8 bands. for a single, stable resonance, one band is enough. for vocals with multiple resonant peaks that shift with pitch, you may need 3-4 dynamic EQ bands. if the problem requires more than 4-5 bands, the resonances are likely too numerous and mobile for band-based tools, and a spectral processor is the better approach.

what is the difference between multiband compression and dynamic EQ?

dynamic EQ applies a narrow boost or cut at a specific frequency when the signal crosses a threshold. multiband compression splits the entire signal into frequency bands and applies broadband compression to each. dynamic EQ is surgical: it targets one frequency. multiband is broad: it affects everything in the band. for isolated resonances, dynamic EQ is almost always the better choice.

when should you give up on EQ and use a resonance suppressor?

when the resonances are numerous (more than 3-4 simultaneous peaks), when they shift in frequency with pitch or performance dynamics, or when fixing one creates another. these are signs that the problem is distributed across the spectrum and moving, which is exactly what spectral resonance suppressors are designed for.

references

a note from the developer

this guide is built on four years of studying psychoacoustics and DSP research. reading papers, building prototypes, making mistakes, and learning from all of it. i am a solo developer in copenhagen, and i am still learning every day.

i built KERN SMOOTH specifically for the problems that EQ and dynamic EQ cannot solve: the shifting, numerous, interacting resonances that require spectral-level resolution. but i also use static EQ and dynamic EQ constantly in my own mixing. the right tool depends on the problem, not on how sophisticated the tool is.

if i got something wrong, missed an approach that works for you, or if you just want to share your workflow for controlling resonance, i genuinely want to hear from you. reach out at jonas@kernaudio.io. every piece of feedback makes these guides better.