Matching tempo and key across beats and vocals

The verse that fits one beat and fights another

You record a verse you love over one beat, then the producer sends a better one and you drop the vocal on top. Every syllable lands a little off the grid, and the hook rubs against the chords on every bar. Or you pull a sample into a session and it clashes with the keys even though both sound fine alone.

The parts disagree about one of two things: when the beats land, or which notes belong.

Why it matters

Tempo and key are fixed in the take

A sung or rapped take carries its tempo and its key with it. Flow, breaths and the gaps between lines sit on the grid it was recorded to, and every note sits on a pitch. Move the take to a new beat and both have to match, or something has to be stretched, transposed or re-recorded. Knowing which moves are clean and which ones damage the take saves sessions.

The science

Two grids, one for time and one for pitch

counts beats per minute, so one beat lasts:

t_{\text{beat}} = \frac{60\,000}{\text{BPM}}\ \text{ms}

At 140 BPM a beat lasts 428.6 ms. Move a verse recorded at 140 onto a 90 BPM beat and every syllable lands in the wrong place; stretching it to fit makes it 56 percent longer. The exception is a tempo exactly half or double. At 70 BPM one beat lasts exactly two beats at 140, so the vocal still lands on the grid and now feels double-time. Trap uses the same relationship: written around 140 with the snare on beat 3, it feels like 70.

Figure 1. One bar at 140 BPM. With the snare on beats 2 and 4 you nod at 140. Move it to beat 3 only and the bar holds two beats at 70 BPM with the snare on the second, a backbeat at half the tempo, so the same grid feels half as fast.

Time-stretching changes length without changing pitch, by cutting the audio into short overlapping frames and laying them out again at a new spacing. Small changes are clean. Larger ones smear and give sustained sounds a hollow, phasey tone, the typical artefacts of these methods (Driedger and Müller, 2016). Work out the change as a ratio: 140 to 145 BPM is 145 / 140 = 1.036, a 3.6 percent stretch.

, the tape method many samplers still use, changes speed and pitch together. Changing the tempo by a ratio moves the pitch by:

n = 12 \log_2 \frac{\text{BPM}_{\text{new}}}{\text{BPM}_{\text{old}}}

One semitone is a ratio of $2^{1/12} \approx 1.0595$ , close to 6 percent. Slowed and pitched-down edits are this effect.

Figure 2. Varispeed ties tempo to pitch. Each semitone multiplies the tempo by 1.0595, so a 140 BPM beat plays at 148.3 BPM one semitone up and at 132.1 BPM one semitone down.

The pitch grid is the key: a home note plus a scale. A minor uses A, B, C, D, E, F and G, and a hook built from those notes fits an A minor beat. The Camelot wheel from Mixed In Key numbers the twelve keys like a clock, A for minor and B for major. Keys with the same number share all seven notes (8A, A minor, and 8B, C major). One step around the wheel is a fifth away and shares six.

Figure 3. Notes each key shares with A minor (8A). The relative major shares all seven and a neighbour on the wheel shares six. A major, the same home note with a major scale, shares four. E♭ minor, opposite on the wheel, shares two.

Hear the grid side of this: the same beat from 60 to 160 BPM. Notice which tempos feel like halves or doubles of each other.

Listen

Change only the tempo

The same pattern and the same sounds from 60 to 160 BPM. Notice where the beat stops feeling heavy and starts feeling urgent.

Loading demo…

Demo volume100%

Start with your speakers or headphones low.

Try it

Match a sample, then set pitch correction

Steps 5 to 7 need a pitch corrector with key and scale settings, which several DAWs include.

Find the sample's tempo. Count the beats in the loop and time it: $\text{BPM} = 60 \times \text{beats} / \text{seconds}$ . A four-bar loop has 16 beats, so if it lasts 6.86 seconds it is at about 140 BPM.
Find its key with your DAW's detection or an online tool, then confirm by playing the root under it. Detection can report the relative major instead of the minor, because the notes are the same.
Stretch it to your session tempo. Try a 3 percent change and a 10 percent change and listen to the drums in the sample for smearing.
Transpose by the smallest move. A sample in B minor needs 2 semitones down for an A minor session. One in C major needs nothing: it already uses every note of A minor.
On your vocal, set the pitch correction key and scale to the beat's, such as C minor.
Sing a line that uses the raised seventh, B natural in C minor. With the plugin on natural minor, listen to it drag that note to a neighbour. Add the note to the scale and listen again.
Set the retune speed by ear: fastest for the hard-tuned effect, slower for gentle correction.

0 of 7 steps done

Common mistake

Trusting the detector and the default scale

You set pitch correction to whatever the key detector reported and leave the scale on natural minor. Minor-key trap and drill melodies often borrow the raised seventh of harmonic minor, G♯ in A minor. A natural minor setting pulls it to G or A, and the line warbles. Check the melody's notes against the scale before you blame the singer.

The second mistake is stretching too far. Past a few percent, re-recording the part at the new tempo usually sounds better than any algorithm. And when a hook sits too high, ask the producer to transpose the beat in the project rather than pitch-shifting the finished file: each semitone multiplies every frequency by 1.0595, so two semitones down takes an 808 on F1 from 43.7 Hz to D♯1 at 38.9 Hz, and the drums would be retuned with it.

Takeaway

Check both grids before you commit

Before you move a part between projects, check the tempo ratio and the key. Exact halves and doubles keep the grid; small stretches are clean; anything bigger wants a new take. Neighbouring keys share six notes, share all seven, and pitch correction needs the scale the melody really uses. More on how correction changes a performance is in the lesson on pitch correction and confidence.

Sources

Show the source

Driedger, J., & Müller, M. (2016). A review of time-scale modification of music signals. Applied Sciences, 6(2), 57.

Matching tempo and key between beats, vocals and samples

In short