choosing your approach

the right tool for the right problem

you have a harsh vocal. you reach for an EQ and cut 3 dB at 3 kHz. it helps, but the next phrase sounds dull because the EQ cut is always on. so you switch to a dynamic EQ. that catches the loud phrases, but the harshness shifts between vowels and the band misses half of them. so you add a second band, then a third, and now you are chasing the problem around the spectrum.

this is what happens when you choose a tool based on habit instead of diagnosis. emergency rooms triage patients: the worst problem gets the simplest effective treatment first. resonance control works the same way. the decision is not about which tool is “best.” it is about which tool matches the problem you actually have.

this guide gives you a framework for that decision.

the four approaches

before choosing, understand what each tool does differently.

static EQ: a fixed gain change at a fixed frequency. always on, always the same. zero latency, near-zero CPU. the simplest tool available.

dynamic EQ: a gain change that activates only when the signal at that frequency exceeds a threshold. you get 4-8 bands to place manually. near-zero latency.

multiband compression: divides the signal into 3-6 broad frequency bands and applies independent compression to each. controls the overall level envelope of each band, not individual resonances within it.

spectral processing: decomposes the signal into hundreds or thousands of frequency bins via FFT. identifies and reduces individual spectral peaks in real time. higher latency (23-93 ms at typical settings), higher CPU, but acts on every frequency simultaneously.

the decision framework

ask three questions about your resonance problem. the answers point to the right tool.

question 1: is the resonance consistent?

yes, always present: the same frequency, the same intensity, on every note and phrase. examples: a room mode, a microphone presence peak, a synth patch with a fixed spectral character. use static EQ. it is the simplest, cheapest, and most transparent solution.

sometimes present: appears on louder phrases, certain vowels, or when the performer pushes harder. absent on quieter moments. use dynamic EQ. the band activates only when the problem shows up.

constantly shifting: moves with pitch, changes with vowel shapes, varies with playing dynamics. use spectral processing. it follows the resonance wherever it goes.

question 2: how many frequencies are affected?

one or two fixed frequencies: a single ringing note, one body resonance. static EQ or dynamic EQ. one or two bands cover it completely.

three to five frequencies: multiple resonances, each needing individual attention. dynamic EQ with careful band placement. this is the limit of what 4-8 bands can handle. beyond this, you are spending more time tuning bands than mixing.

many frequencies, or the number changes: dense resonances across a range, new peaks appearing on different notes. spectral processing. it acts on every frequency simultaneously without requiring you to predict where problems will appear.

question 3: what is the source?

single instrument or vocal: you can be precise and aggressive. any tool works depending on the answers to questions 1 and 2.

drum bus or group: dynamic EQ or gentle spectral processing. transient preservation matters here, so avoid slow-responding tools or heavy settings that soften the attack.

mix bus or master: gentle spectral processing or mid/side dynamic EQ. sensitivity must be low. everything on the bus is affected simultaneously. see our guide to mix bus resonance for detailed settings.

when static EQ is the right answer

static EQ gets a bad reputation because it is simple. but simplicity is a feature when the problem is simple.

use it for:

• room modes: a 6 dB resonance at 280 Hz from a boxy room. always there, always the same frequency. a narrow notch (Q 6-10) at -4 to -6 dB fixes it permanently
• microphone character: a condenser with a 3 kHz presence peak that sounds harsh on every take. a gentle bell cut (Q 3-5, -2 to -3 dB) tames it
• fixed synth patches: a pad or lead with a static spectral resonance. if the patch does not change, neither should the fix
• post-saturation harshness: saturation generates consistent high-frequency harmonics. a shelf or tilt after the saturator handles this cleanly

do not use it when the resonance is intermittent or moves. a static cut is always on, always reducing. on moments where the resonance is not present, the cut removes useful energy and makes the source sound duller.

when dynamic EQ is the right answer

dynamic EQ is the most versatile tool for individual tracks. it acts only when needed and returns to unity when the problem is absent.

use it for:

• vocal harshness on loud phrases: the singer pushes into the mic on certain words. a dynamic EQ band at 3 kHz with a moderate threshold catches the peaks and leaves the quiet phrases alone
• guitar pick attack: the plectrum excites a resonance at 4-5 kHz on aggressive strums but not on fingerpicked passages. a fast-attack dynamic band handles the transient
• intermittent cymbal ring: a specific cymbal has a ringing overtone that appears on hard hits. a dynamic band can catch it without dulling the cymbal on softer playing

the limitation is scale. if you need more than 3-4 active bands, you are either chasing a problem that is too complex for manual placement or your bands are stepping on each other. at that point, consider spectral processing.

when multiband compression is the right answer

multiband compression is not a resonance tool. it is a dynamics tool that operates per frequency band. but it overlaps with resonance control in specific scenarios.

use it for:

• broadband level control: the low-mids of a drum bus get boomy on heavy sections. a multiband compressor tames the overall level of the 200-500 Hz band without touching individual resonances within it
• mastering tonal balance: the high end of a master gets too bright on choruses. a multiband compressor on the 8-15 kHz band controls the overall brightness without acting on specific frequencies
• dynamic tonal shaping: you want the bass to be more consistent but the midrange to breathe. per-band compression gives you independent dynamic control

do not use it for narrow resonances. a multiband compressor’s bands are broad (typically 1-2 octaves). a resonance at 3 kHz sits inside a band that also contains 2-4 kHz. compressing the entire band affects everything in that range, not just the resonance.

when spectral processing is the right answer

spectral processing is the heaviest tool in terms of CPU and latency, but it solves problems that no other approach can.

use it for:

• shifting resonances: a vocalist whose harshness moves between 2.5 and 4 kHz depending on vowel shape and pitch. spectral processing follows it automatically
• dense resonance clusters: a bowed string instrument with multiple simultaneous resonances that shift with each note. no reasonable number of dynamic EQ bands can cover this
• cymbal and percussion harshness: inharmonic overtones spread across the entire upper spectrum. spectral processing acts on all of them at once
• mix bus finishing: residual harshness from spectral energy accumulation across many tracks (see question 3 above)
• mastering: treating a finished stereo mix where you cannot access individual sources

combining approaches

the best vocal chains often use multiple tools, each at conservative settings:

1. static EQ: notch the room resonance (fixed, always on)
2. dynamic EQ: catch the 1-2 worst harshness frequencies (intermittent)
3. spectral processing: gentle pass for everything else (shifting resonances, overall smoothing)

this layered approach keeps each tool at moderate settings. no single processor is doing heavy lifting, which minimizes artifacts from any one stage.

the quick reference

problem type	tool	why
fixed room mode	static EQ	always there, never moves
mic presence peak	static EQ	consistent across takes
loud phrase harshness	dynamic EQ	intermittent, 1-2 frequencies
shifting vocal resonance	spectral	follows vowels and pitch
cymbal overtones	spectral	dense, inharmonic, spread across spectrum
boomy drum bus	multiband comp	broad level control, not per-resonance
mix bus harshness	gentle spectral	accumulated energy from many sources
mastering harshness	M/S spectral	targets center (vocal) without affecting sides

frequently asked questions

frequently asked questions

when should you use a static EQ instead of a dynamic EQ for resonance?

use static EQ when the resonance is consistent and predictable: a room mode that colors every take, a microphone with a fixed presence peak, or a synth patch with the same spectral character on every note. if the resonance does not move or change intensity, a static EQ is the simplest and most CPU-efficient solution.

what is the difference between multiband compression and a resonance suppressor?

multiband compression divides the signal into 3-6 broad frequency bands and controls the overall level of each band. a resonance suppressor analyzes hundreds or thousands of frequency points and targets only the specific peaks that protrude above the local average. multiband is a level tool. a resonance suppressor is a spectral precision tool.

does spectral processing add latency?

yes. spectral processing requires an FFT analysis window, which introduces latency proportional to the window size. typical values: 23-93 ms depending on the FFT size and sample rate. static EQ and dynamic EQ have zero or near-zero latency. this makes spectral processing less suitable for live monitoring but fine for mixing and mastering where latency is compensated automatically.

can you use multiple approaches together?

yes, and this is often the best strategy. use a static EQ notch for a fixed room resonance. add a dynamic EQ band for an intermittent harshness. use a gentle spectral processor on the bus for residual build-up. each tool handles a different type of problem, and combining them keeps each one at conservative settings.

which approach is best for vocals?

it depends on the vocal. if the harshness is on specific notes or phrases, a dynamic EQ works well. if the harshness shifts with vowel changes or pitch, spectral processing follows it automatically. most vocal chains benefit from a dynamic EQ at 1-2 specific frequencies plus gentle spectral processing for everything else.

references

a note from the developer

the hardest lesson i learned building SMOOTH was that my own tool is not always the right tool. i genuinely believed spectral processing was the answer to everything. it is not. a static EQ notch at a room resonance is faster, lighter, and more transparent than a full spectral pipeline aimed at the same problem.

i still use static EQ on almost every session. i use dynamic EQ more than i use SMOOTH. spectral processing earns its place when the problem is complex enough to need it: shifting resonances, dense overtone clusters, mix bus accumulation. for simple problems, simple tools win. i would rather you use the right tool and never buy SMOOTH than use SMOOTH on a problem it was not designed for.

if your resonance workflow looks completely different from what i described, i am curious. jonas@kernaudio.io.