Skip to content

Resonance can sing or wreck the mix

Resonances give instruments their voice and turn harsh only when they ring too long or too loud. Learn what Q means and how to cut only what hurts.

Updated · 6 min read Lesson 8 of 14 in Audio Science

In short

  1. Resonances give instruments and voices their character. They are a problem only when they are too strong for the context.
  2. Q is centre frequency divided by bandwidth, and the narrower a resonance is, the longer it rings.
  3. Sweep with about 4 dB of boost, cut 2 to 3 dB, and judge every cut with the full mix playing.
In this lesson (8)

The sweep that empties a vocal

A vocal sounds slightly harsh, so you create a narrow band, boost it by 15 dB and sweep it through the upper mids. Everywhere you stop, something whistles, so you turn each spot into a deep notch. Four notches later the harshness is gone, and so is the voice. In the mix it sounds thin and far away.

A 15 dB narrow boost makes almost any frequency sound like a problem. Cut everything that rings under that much boost and you strip out the that make an instrument sound like itself.

Why it matters

Resonances are the instrument

Most acoustic sound is shaped by resonance. A snare's shell and heads ring at a few frequencies, usually with a somewhere in the low hundreds of hertz, depending on tuning. A guitar body boosts some notes more than others. A voice is a buzz from the vocal folds filtered by the throat, mouth and nose, and the resonances of those spaces, the , are what make an "ah" different from an "ee".

A resonance becomes a problem when it is too strong for its context: a narrow peak that pokes out on loud notes, or a ring that keeps sounding after the note has stopped and smears the next one. The job is to tell those apart from the resonances that carry the tone, and that starts with knowing what makes a resonance narrow or broad.

The science

Q, bandwidth and ringing

A resonant system stores energy at its natural frequency and gives it back slowly. How narrow the resonance is and how long it rings are two views of one property. The QQ relates the centre frequency fcf_c to the bandwidth BB, measured between the points 3 dB below the peak:

Q=fcBQ = \frac{f_c}{B}

A resonance at 1 kHz with a bandwidth of 100 Hz has a Q of 10. The same number sets the ringing. For a simple resonance, the envelope after the input stops decays as e−πBte^{-\pi B t}, so the time to fall by 60 dB is about

T60≈2.2B=2.2 QfcT_{60} \approx \frac{2.2}{B} = \frac{2.2\,Q}{f_c}

A 1 kHz resonance at Q 5 dies away in about 11 ms. At Q 25 it rings for about 55 ms. Narrow means long.

Figure 1. Two 1 kHz resonances struck once, drawn over 20 ms. At Q 5 the ring falls 60 dB in about 11 ms. At Q 25 it is still only about 22 dB down after 20 ms.

Filters use the same physics. In a digital filter, resonance comes from feedback: part of each output is fed back into the next calculation. Raise the resonance control on a synth and the feedback lifts a peak at the cutoff. For a two-pole low-pass, the gain at the cutoff equals Q, so Q 4 is a 12 dB peak and Q 10 a 20 dB peak, and the filter rings at that frequency on every .

Figure 2. A two-pole low-pass at 1 kHz with three resonance settings. At Q 0.7 there is no peak, only a bend 3 dB down at the cutoff. At Q 4 the cutoff is boosted by 12 dB, and at Q 10 by 20 dB.

Listen

Shape a sound with a filter

Chords and soft noise, or a real mix, through one filter, with the spectrum drawn live. Sweep the cutoff and raise the resonance.

Loading demo…
100%

Start with your speakers or headphones low.

Q on an EQ plugin is a guide, not a standard. Manufacturers define a bell's bandwidth slightly differently, especially for boosts against cuts, so the same number can give different widths in two EQs. Trust the curve and your ears over the number.

Try it

The quiet sweep

0 of 7 steps done

Real problem resonances stand out even at a 4 dB boost and usually need only a few dB of cut. Most of what screamed under a 15 dB boost was the normal tone of the instrument.

Common mistake

Notching every peak

You put a row of narrow cuts across the midrange of every vocal, often placed by eye from an analyzer. The peaks they remove include the formants that carry the vowels and the singer's character. The vocal gets smoother, harder to understand and less able to sit on top of the track. Formants also move with every vowel, so a fixed notch on one of them is in the wrong place most of the time.

The other mistake is judging resonance in solo. A ring that is obvious alone can be covered by the band, and a peak that is fine alone can clash with another part at the same frequency. Decide in context.

Takeaway

Cut what hurts, keep the voice

Hunt resonances with small boosts, fix them with small cuts and judge them with the mix playing. For a resonance that only bites at loud moments, such as a vocal that turns harsh on held high notes, use a band. It cuts only when that frequency crosses a threshold and leaves the quiet passages alone. A track with a few controlled peaks still sounds played. A track with every peak removed sounds like nobody is playing it.

3 questions

Check yourself

  1. An EQ band at 2 kHz has a Q of 4. What is its bandwidth?
  2. Two resonances sit at the same frequency, one at Q 5 and one at Q 25. What does the second one do differently?
  3. Why sweep for resonances with a small boost instead of a large one?

Sources

Show all 2

Next lesson · 9 of 14

Phase is timing with consequencesTwo copies of a sound add up or cancel depending on their timing. Fix polarity and alignment on related mics before you reach for EQ, then check the mix in mono.

Previous: Filters are shape machines

Portrait of Virzy Guns

Written by Virzy Guns

Songwriter and producer, founder of Virzy Guns Production. Now building HealingWave.