How to align vocals, and how tight is too tight

You stack a lead with its double and something feels off. Not wrong, exactly — just soft around the edges. The words smear, the “s” at the end of a line hisses twice, and a chorus that sounded huge in your head sounds like two people who almost agree. That’s timing, and it’s one of the most common reasons a home vocal sounds less finished than a record.

This guide covers what alignment actually fixes, how much timing difference you can hear, a fix order that saves hours, how to do it in every major DAW, and how to know when to stop.

What “aligned” means

Aligning vocals means making the timing of one take follow another. You pick a guide — almost always the lead vocal — and move the other takes so their syllables start and end where the guide’s do. Pitch, tone and vibrato stay as they were sung. That is the whole point: you keep two different performances, and only take away the timing differences that read as mistakes.

Three moments in every syllable matter, in this order:

1. The start. The first consonant or the attack of the vowel. This is where the ear decides whether two voices are together.
2. Hard consonants inside the word. A “t”, “k”, “p” or “ch” that lands twice sounds like a flam on a snare.
3. The release. Where the note ends. One voice hanging on after the other is very exposed at the end of a phrase.

The middle of a long vowel matters least. Two singers holding “oh” slightly differently is exactly what makes a double sound thick.

How much difference you can hear

Two sounds that arrive close together are heard as one: the ear fuses them. As the gap grows, the second voice stops adding body and starts adding blur, then a distinct second attack, and eventually an echo. Where those thresholds sit depends on the sound — a breathy vowel hides a lot, a crisp “t” hides almost nothing — but a rough map helps you decide what is worth fixing:

Two consequences that change how you work:

• You don’t need zero. Getting syllable starts within roughly 5–10 ms is tight for almost any style. Chasing the last millisecond adds artifacts and makes the stack sound like a copy.
• Sibilants are the giveaway. Two “s” sounds that overlap add up in level — two similar, unrelated sounds sum about 3 dB louder — and two that don’t overlap sound like one long hiss. Either way, sibilance is where misalignment shows first.

Before you start

Ten minutes of preparation saves an hour of nudging.

1. Comp the lead first. The guide has to be final. If you align to a lead you later re-comp, you align again.
2. Tune the lead, if it needs it, before aligning anything to it. Pitch correction tools like Melodyne or Auto-Tune don’t move timing on their own, but a re-sung or re-comped word does.
3. Consolidate each take from the same start point. One continuous file per part, starting at the same bar, makes everything easier — and it is what alignment plugins need.
4. Gain-match roughly. A double 10 dB quieter than the lead is hard to judge visually and by ear. Bring it up while you work, set the final level later.
5. Keep the original. Duplicate the double track and mute the copy. You will want to A/B.

The order of fixes

Work from big to small. Each step makes the next one smaller.

1. Whole-phrase offset

Many doubles are simply late or early by the same amount for a whole phrase — the singer reacted to the lead in their headphones. Slide the whole region first. If every syllable is 30 ms late, one nudge fixes thirty syllables.

2. Word starts

Now go word by word. Line up the first consonant of each word with the lead. This alone takes most stacks from sloppy to tight.

3. Hard consonants

Listen for any “t”, “k”, “p”, “d” or “ch” that still clicks twice. Those sit inside words and are easy to miss while you’re fixing starts.

4. Phrase ends

Match where each phrase stops. Shortening a double’s tail is easier and cleaner than stretching it: a short fade at the lead’s release point often does the job.

5. Leave the rest

If it sounds right after step four, stop. The differences that remain are the reason you recorded a second take.

Doing it in your DAW

There are two manual techniques, and good editors use both. Cut and move: split the double just before a word, slide the piece, crossfade the cut. It sounds perfectly clean because nothing is processed, but it only moves starts. Time-stretch: put a marker on a syllable and drag it, and the audio around it stretches or squeezes. That changes lengths too, at the cost of possible artifacts.

Consonants and breaths

Experienced vocal editors don’t only align consonants — they often remove them from the doubles. The lead carries the “s”, “t” and the end of the word; the doubles carry the vowels.

• Ending “s” and “t”: fade the double out just before them, or cut them and let the lead finish alone. Nobody will hear that the double is missing a consonant. Everybody hears two.
• Starting consonants: keep them on the double only if they line up within a few milliseconds. Otherwise, start the double at the vowel.
• De-essing: de-ess the doubles harder than the lead. Their job is body, not air.
• Breaths: cut or fade them on the doubles. One breath sounds human; three slightly offset breaths sound like noise.

How tight, part by part

Not every part should be equally tight. A useful starting point:

Style matters too. Modern pop and rap usually want doubles very tight. Rock and indie often leave them looser on purpose. In R&B, align the start and end of a run, not every note inside it — runs never match note for note, and forcing them to sounds robotic.

Mistakes that cost you

• Aligning to the wrong take. Every part follows one guide. Aligning a harmony to a double that was aligned to the lead stacks up small errors.
• Too tight. Two very similar takes lined up within a millisecond or two start cancelling each other at some frequencies — comb filtering — and the vocal goes hollow and phasey. If that happens, back off.
• Stretching too far. If a syllable needs a big move, cut and slide it instead.
• Clicks at the edits. Every cut needs a crossfade, or at least a short fade at a quiet point.
• Aligning what should be muted. If a word on the double is badly sung, muting it is better than aligning it.
• Judging with your eyes. Waveforms that look lined up can still flam, and ones that look off can sound perfect. The waveform tells you where to look; your ears decide.

Checking your work

1. Solo the lead and one double and listen at low volume. Low volume makes flams easier to hear.
2. Check in mono. Panned doubles hide timing problems that a mono speaker — or a phone — will reveal.
3. A/B with the muted original. Aligned should sound tighter, not more processed.
4. Listen in the full mix. Some differences you fixed are inaudible once the track plays. Some you left will suddenly stick out.

With a vocal alignment plugin

A vocal alignment plugin does the matching for you. It analyses both takes, finds which part of the double belongs to which part of the lead, and builds a time map — a continuous version of the markers you would place by hand — then stretches the double along it while keeping its pitch. A three-minute song with four doubles goes from an afternoon to a few minutes.

What to check afterwards, whichever tool you use:

• Words that don’t exist in the guide — ad-libs, a different lyric — have nothing to align to. A good tool leaves them alone or tells you about them; a bad one squeezes them somewhere.
• Stretch artifacts on long moves. Listen to the busiest phrases solo.
• Tightness. Most tools have a strength control. Full strength is rarely the best-sounding setting for doubles.

We built Ordinary Align around those three points. You drop in the lead and the double; a Tightness control decides how close is close enough, with about 85% as the default for doubles; Max shift caps how far any syllable can move; and anything in the double that isn’t in the lead is marked Not in guide so you can leave it as sung. It only moves timing, never pitch. If you’re comparing tools, we also wrote an honest guide to VocALign alternatives.

FAQ