Skip to content

Too many layers make sounds smaller

Similar layers do not simply add. Offsets cut notches, detune makes the level swell and sag, and attacks smear. Layer by role instead of by copy.

Updated · 6 min read Lesson 8 of 10 in Sound Design

In short

  1. Copies of a sound add or cancel depending on their phase: double in phase, silence at 180 degrees.
  2. A time offset cuts evenly spaced notches, detune makes the level swell and sag, and mixed attacks smear the start of each note.
  3. Layer by role, with each layer filtered to its own band, and keep the single best layer if the stack does not beat it.
In this lesson (8)

The shrinking wall of sound

You want a massive lead. You load a bright saw preset. It sounds good, but you want huge, so you duplicate the track and load another saw on the copy, then a third. Three leads now play the same notes. You hit play expecting a wall of sound.

Instead the lead sounds smaller: thinner, blurrier, missing the midrange weight it had. Mute two of the layers and the one left sounds heavier and clearer.

Similar waves do not simply add. Whether they reinforce or cancel depends on their , and in a stack of near-identical sounds that relationship shifts from moment to moment and from one frequency to the next.

Why it matters

Copies interfere

Two copies of the same sound in phase add to twice the amplitude, 6 dB louder. Flip the of one and they cancel completely. Most stacks sit somewhere between those extremes, and they keep moving.

Stacks rarely line up exactly. Samples start a few milliseconds apart, start at different points in their cycle, and unison keeps the copies drifting. Each of those does something specific. A fixed time offset between two copies is a : it cuts deep notches at evenly spaced frequencies. A small detune makes the copies drift in and out of phase, so their sum swells and sags at a rate equal to the difference in pitch. Different attacks smear the start of each note.

The stack gets louder on average, because unrelated layers add in power, but it gets less defined. Turn it down to the level of the single layer and it sounds smaller.

Figure 1. Two equal copies of a sine at three phase offsets, computed. In phase they double. A third of a cycle apart, the sum is no louder than one copy. Half a cycle apart, they cancel completely.

Listen

Hear two copies of a bass cancel

The same bass note on two layers. Delay one by a few milliseconds or flip its polarity and the low end thins out, then disappears.

Loading demo…
100%

Start with your speakers or headphones low.

The science

Sums of sines

Two sine waves of the same frequency, with amplitudes A1A_1 and A2A_2 and a phase difference Δϕ\Delta\phi, add to one sine with amplitude:

Atotal=A12+A22+2A1A2cos⁡ΔϕA_{\text{total}} = \sqrt{A_1^2 + A_2^2 + 2 A_1 A_2 \cos \Delta\phi}

With equal amplitudes AA this becomes 2A∣cos⁡(Δϕ/2)∣2A \left| \cos(\Delta\phi / 2) \right|: double at 0°, about 1.41 times (+3 dB) at 90°, the same as one copy at 120°, and silence at 180°.

A time offset Δt\Delta t turns into a phase difference that grows with frequency, Δϕ=360°⋅f Δt\Delta\phi = 360° \cdot f \, \Delta t. The copies cancel wherever that reaches an odd multiple of 180°, which gives the notches of the comb (Smith, Introduction to Digital Filters):

fk=2k+12 Δt,k=0,1,2,…f_k = \frac{2k + 1}{2 \, \Delta t}, \quad k = 0, 1, 2, \dots

An offset of 1 ms puts notches at 500 Hz, 1.5 kHz, 2.5 kHz and up. An offset of 5 ms, easy to get from two samples that do not start together, puts them at 100, 300 and 500 Hz, right in the body of the sound.

Detune works over time instead. Two tones at ff and f+Δff + \Delta f beat Δf\Delta f times per second (Moore, 2012). A 5-cent detune at 110 Hz is about 0.32 Hz apart, a slow swell every three seconds or so. At 880 Hz the same 5 cents is about 2.5 Hz, a fast shimmer. That is why unison sounds lush on top and unstable in the low end.

Figure 2. Two equal tones slightly apart in pitch, computed. Their sum swells to double and falls to silence once per cycle of the pitch difference. On a detuned bass this is a slow rise and fall in level.

Try it

The layered sound role test

Hear how copies interfere, then build a stack that does not.

0 of 7 steps done

The copies sound hollow, swell and smear at the start of each note. The role stack keeps a steady body, a clear attack and air on top, and it survives mono.

Common mistake

Stacking similar presets

Layering is not adding more of the same. Three presets from the same library, in the same octave with the same envelope, fight over one band and interfere with each other there. If two layers do the same job, one of them is only adding level and phase trouble.

The low end is the worst place for it. Slow beating between detuned copies turns into a level that rises and falls under the whole mix, and wide unison can partly cancel when the mix is folded to mono. Keep the sub as one clean source.

Takeaway

Layer by role

Give every layer one job: body, edge, width or air. Filter each so its band overlaps the others as little as possible, give the edge layer the sharpest attack and the body the steadiest pitch, and keep the lowest layer in mono.

Figure 3. A stack layered by role, sketched. Each layer owns its own band, so together they add body, bite and air instead of interfering in one range.

Then check the stack against its single best layer at matched loudness. If the stack does not win, use the single layer.

3 questions

Check yourself

  1. Two equal copies of a sound are 180 degrees out of phase. What do you hear?
  2. A copy of a lead is delayed by 1 ms. Where is the first notch?
  3. Why does a 5-cent detune sound like a slow swell on a bass note but a shimmer on a high note?

Sources

Show all 2
  • Moore, B. C. J. (2012). An Introduction to the Psychology of Hearing (6th ed.). Emerald.
  • Smith, J. O. Introduction to Digital Filters with Audio Applications. CCRMA, Stanford University. ccrma.stanford.edu/~jos/filters/

Next lesson · 9 of 10

Sonic signature beats another layerFive stacked presets make a hook louder, not easier to recognize. Layers that start together fuse into one blend, so build the signature into one sound.

Previous: Small speakers need bass harmonics

Portrait of Virzy Guns

Written by Virzy Guns

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