How to find the key of a vocal or a cappella

Why vocals are a special case

Key and tempo fail in completely different ways on a vocal, and knowing which is which saves a lot of time.

Key works. It reads pitch content, and a sung melody is pitch content. It needs no beat, no kick and no transient. An a cappella is a perfectly ordinary input.

Tempo mostly does not. It reads transients, and sung transients are soft. Worse, plenty of vocal takes have no reliable pulse in them at all — the singer was following a groove that is not in the file.

So: trust the key on a vocal. Be sceptical of the tempo. A tool that returns a key and declines to print a BPM on an a cappella is behaving correctly, not failing.

The fastest route

If the vocal is a file, drop it in. DropWave gives you the key, the Camelot code, the measured tuning offset and a confidence band from one drop.

If the vocal is a channel in your session — which it usually is — you do not need to bounce it at all. Insert the plugin on the vocal bus, play, press Analyze, read the key. That is the whole thing.

The check that matters: major or minor

The most common wrong answer on a vocal is the relative key — A minor reported where the vocal is really C major, or the reverse. They contain identical notes, and a melody that never lands on its tonic gives nothing to separate them with.

Vocals are unusually prone to this, because a sung line often orbits the third or the fifth rather than the root.

To check:

1. Play a sustained chord of the reported key under the vocal.
2. Play the relative — same number, other letter on the Camelot wheel; 8A ↔ 8B.
3. One of them will feel like home under the phrase endings. That is your answer.

Fifteen seconds, and it settles the one ambiguity no tool can settle for you.

When the vocal is detuned

Vocals arrive off concert pitch more often than most material — they come as bounces from sessions that were not at 440, or as rips, or already pitched to fit something else.

A tool that assumes standard tuning will land on the neighbouring key. One that measures the offset will not, and will tell you how far off the file is — which is exactly what you need before you write anything under it, because you will have to match it.

Getting the tempo anyway

If you genuinely need it, do not expect a detector to find it:

1. Find one clear consonant — a "t", a "k", a plosive — at the start of a phrase.
2. Find the same syllable four or eight bars later.
3. Measure the distance in seconds and divide: BPM = beats × 60 ÷ seconds.
4. Nudge until the loop holds over the full length.

Or more practically: warp it. Every DAW has a mode built for this, and a vocal with a hand-placed grid is worth more than a vocal with a number attached.

Checking it against your session

1. Get the Camelot code. If the vocal is 8A and your session is 8A, 7A, 9A or 8B, you are fine.
2. Further away? Pitch it. Each semitone moves 7 positions — so +1 semitone takes 8A to 3A.
3. Watch the formants. Past about ±2 semitones a straight shift starts to sound like a different person. Use a formant-preserving mode if you need more.
4. Check the tuning offset. If the vocal is 30 cents flat and your session is not, no amount of correct key matching will make it sit right.

Related reading

• How to find the key of a sample
• How pitch-shifting changes a sample's key
• The plugin — reading a vocal channel without bouncing it