Skip to content

Why a vocal double works best unnoticed

A double should make the lead feel bigger without sounding like a second singer. Tuck it under the lead, filter it, and fix the consonants that give it away.

Updated · 5 min read Lesson 10 of 12 in Vocal Production

In short

  1. The brain groups a tight double with the lead into one voice, and the small differences between the takes are heard as thickness.
  2. The illusion breaks when the double is nearly as loud as the lead or when its hard consonants land at a clearly different moment.
  3. Find the level where you first hear a second singer, then pull the double back from there.
In this lesson (8)

Two singers where you wanted one big voice

You record a double of the chorus to make it bigger. Then you push its fader until you can clearly hear it, because that seems to be the point of recording it. Now the chorus sounds like two singers standing side by side, and the lead has lost its focus.

A double is not meant to be heard as a second performance. It works when the lead feels thicker and wider and the listener cannot point to the reason.

Why it matters

A loud double blurs the lead

No two takes line up exactly. Overlapping vowels blend well enough. Hard consonants and s sounds do not: when the double is loud, each t and s arrives twice, a moment apart, and the words get a smeared edge.

The centre of the mix also changes. It now holds two voices at similar levels, and the listener's attention has no clear place to land. Turning the double up for more width makes this worse, because the extra width comes with a second set of words.

The science

Grouping, not echo

The brain decides what belongs to one sound source using cues such as parts that start together, move in pitch together, share a and come from the same place (Bregman, 1990). A tight double matches the lead on almost all of them: same singer, same notes, nearly the same timing. So the brain groups the two takes into one voice. The small differences that remain, a slightly different pitch path or a slightly different attack, do not split the voice in two. They make it sound thicker, the way a section of violins sounds fuller than one player.

The grouping breaks in two ways. If the double is nearly as loud as the lead, it competes as a second voice. If an onset drifts too far, especially on a hard consonant, the brain hears two events.

Figure 1. A hard consonant in the lead and in the double. When the double lands close to the lead, the two fuse into one attack. When it lands clearly later, you hear a second click, a flam, and the double gives itself away.

Doubling is often explained with the , but that describes something else: a sound and its own delayed copy, which the ear fuses and places where the first arrival came from. Copy the lead and delay it and you get exactly that, plus . A fixed delay τ\tau cancels every frequency at which the copy arrives half a cycle late:

fnotch=2k+12τ,k=0,1,2,…f_{\text{notch}} = \frac{2k + 1}{2\tau}, \quad k = 0, 1, 2, \ldots

With a 10 ms delay the notches fall at 50, 150 and 250 Hz and on up every 100 Hz. A real double is a second performance with no fixed delay, so no fixed comb pattern forms. Its differences keep changing, which is why it sounds bigger than a copy.

Try it

Find the level where the double disappears

0 of 6 steps done

Muted, the chorus should feel smaller and thinner. Unmuted, it should feel fuller while you still hear one singer. If the t and s sounds flam, fix the timing of those consonants in the double before you touch the level again.

Common mistake

A full-range, unedited double

A double left at full range adds its low end to the lead's and thickens the low mids into mud. Its s sounds arrive at slightly different moments from the lead's and flam. You do not need either from the double. Filter it as in the experiment, de-ess it harder than the lead, or cut its s sounds out completely and let the lead carry them. Line up the hard consonants by hand or with an alignment tool.

Panned pairs need the same care. Once a double moves out of the centre it is no longer hidden behind the lead, so it has to sit lower to stay unnoticed.

Figure 2. Two common places for doubles: one tucked straight behind the lead, or a pair panned wide. Either way they sit further back and lower than the lead, so the lead stays the voice the listener follows.

Takeaway

Felt, not heard

Treat a double as part of the lead's sound, not as a second part. Filter it, fix its consonants and set its level by the mute test: you should miss it when it goes, without hearing it while it plays.

3 questions

Check yourself

  1. Why does a real double sound thicker than a delayed copy of the lead?
  2. You copy the lead and delay it by 10 ms. Summed at equal level, where is the lowest notch?
  3. At the level you chose, the t and s sounds of the chorus flam. What do you fix first?

Sources

Show the source
  • Bregman, A. S. (1990). Auditory Scene Analysis: The Perceptual Organization of Sound. MIT Press.

Next lesson · 11 of 12

Ad-libs are arrangement movesAd-libs that sing over the lead compete with it for attention. Put them in the gaps, away from the centre, and they push the song forward instead.

Previous: Fast retune speed erases the movement between notes

Portrait of Virzy Guns

Written by Virzy Guns

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