Vocal harmonies and doubles: how to record and stack them so the lead still leads

A lead vocal tells the story. Doubles and harmonies decide how big, how intimate and how polished that story feels. Get them right and the chorus lifts without anyone noticing why. Get them wrong and the vocal turns into a blurry choir that buries the one voice people came to hear.

This guide sits between recording vocals at home and comping and timing edits. That guide covers tightening doubles in a sentence or two; this one is about the whole stack: what to record, how to record it, and how to place it in the mix.

Doubles and harmonies are different jobs

It helps to name the parts before you record them.

• A double is the same melody, sung again. Two real performances never match exactly, and those tiny differences in pitch and timing are what make a double sound thicker than one voice.
• A harmony is a different melody that moves with the lead: a third above, a fifth below, whatever the chords support. It adds colour and, often, emotion.
• A stack is several of either, layered: two doubles of the lead, plus two takes of a high harmony, plus two of a low one.

Doubles add weight. Harmonies add meaning. Most bedroom mixes reach for both on every line, which is how a verse ends up sounding like a chorus and the chorus has nowhere to go.

Step 1: Plan the stack before anyone sings

Write the arrangement first. A simple plan works for most pop, indie and R&B songs:

1. Verse: lead alone, or lead plus one quiet double on the last line of each phrase.
2. Pre-chorus: add one harmony, often a third above.
3. Chorus: lead, two doubles, and a harmony pair (one above, one below), each recorded twice.
4. Final chorus: the same stack, plus one new part (an octave up, or an ad-lib) so it still feels like a step up.

If you are not sure which harmony notes fit, sing or play the chord under each line and pick the nearest chord tone above the lead note. Our guide to chord progressions for producers covers the theory you need. Record a rough guide of each harmony on a keyboard so the singer has something to match, then mute it.

Step 2: Record the doubles as real performances

The double has to match the lead closely enough to fuse but not so closely that it cancels into a phasey copy. That means singing it again, not copying the lead track and nudging it.

• Keep the setup identical. Same mic, position and input gain as the lead, so the tones blend.
• Listen to the lead in the headphones while singing. The singer matches phrasing, word lengths and where breaths fall.
• Sing slightly less. Doubles often sound better delivered a touch softer, with less vibrato and fewer runs. Leave the flourishes to the lead.
• Loop the section. In Ableton Live, the manual notes that take lanes are created automatically when you record new clips in the Arrangement, so you can loop a chorus and collect several doubles in one go. Logic Pro does the same with take folders, described in its comping overview.

Comp the doubles from those takes the same way you comp the lead, but choose for match, not magic. The best double is the one that disappears into the lead.

Step 3: Record harmonies in pairs

A single harmony track sounds like a second singer. Two takes of the same harmony, panned apart, sound like a texture. That is usually what you want.

• Record each harmony line at least twice.
• Sing harmonies quieter and more evenly than the lead, and keep consonants soft.
• If the singer struggles with a part, record it slowly, one phrase at a time. There is no prize for one-take harmonies.
• Low harmonies need less level than you think. A third below can muddy the mid-range fast.

Step 4: Tighten where the ear notices

Stacks fall apart at word starts and, above all, word endings. Two voices hitting "s" or "t" at slightly different moments sounds like a hiss or a flam.

1. Line up the start of each phrase with the lead.
2. Trim the ends of the doubles and harmonies so they stop with, or just before, the lead.
3. Cut or heavily reduce sibilants and hard consonants on the backing parts, and let the lead carry them. A de-esser on the backing bus helps.
4. Leave the middles of words alone unless something is clearly out. That small looseness is the point of a double.

In Live, warp markers let you move audio in time; the warping chapter covers the mechanics.

Step 5: Pitch, lightly

Fix only the notes that rub against the lead. Harmonies that are perfectly tuned and perfectly timed tend to sound like a synth pad. If you use pitch correction, apply it to the backing parts more gently than to the lead, or not at all.

Step 6: Pan, level and bus

A starting layout that works for many songs:

• Lead: centre.
• Doubles: either one double in the centre, tucked 8 to 12 dB under the lead, or two doubles panned moderately left and right.
• Harmony pairs: panned wider than the doubles, take one left, take two right.
• Everything except the lead: routed to one backing vocal bus, so you can EQ, compress and ride the whole stack with one fader. Our bus processing guide has the routing.

On the bus, high-pass more than you would on the lead, cut some low mids, and add a little more reverb than the lead gets so the stack sits behind it. See stereo width and panning for the trade-offs of going very wide.

The test: solo the lead, then bring in the bus. The vocal should sound bigger, not different. If the words get harder to understand, the stack is too loud.

When fake harmonies are fine

Pitch-shifting a copy of the lead is a legitimate tool, especially for a quick demo or a deliberate effect. The catch is formants, the resonances that make a voice sound like a particular person. Shift pitch without protecting them and voices go chipmunk or giant.

Ableton's warping chapter is clear on this: in Complex Pro mode, a Formants setting of 100% keeps the original formants "even if the pitch is changed significantly", and the control does nothing if the clip is not transposed. Other DAWs and pitch tools have similar formant controls.

Even with formants preserved, a shifted copy has the same timing and phrasing as the lead, so it fuses too perfectly and loses the life a second performance brings. Use it for a sketch, then re-sing the parts that matter.

What we think

Backing vocals are where a lot of home producers hide. The stack gets bigger every time the lead feels exposed, until the singer is no longer really heard. That is a production choice dressed up as polish.

Our view is that a stack is something the lead earns. Start every song with the lead alone and only add a part when you can say what it is for: weight on the hook, a harmony that changes how a line feels, a call-and-response that tells part of the story. If you cannot name the job, mute it.

We also think re-singing beats generating. Pitch-shifted and AI-generated harmonies are getting good, and they have a place in demos. But a real second take carries the singer's breath and hesitation, and that is often what makes a chorus feel human. Spend the extra twenty minutes in the booth. Your voice, stacked by you, is part of your sound in a way a preset never will be.

Quick checklist

1. Plan the stack by section before recording.
2. Record doubles as fresh performances with the lead in the headphones.
3. Record each harmony at least twice.
4. Tighten word starts and ends; trim sibilants on backing parts.
5. Tune the backing parts lightly.
6. Pan doubles moderately and harmonies wider; route all backing parts to one bus.
7. Check that the words are clearer, not murkier, with the stack in.