← Articles

Metric Modulation in Composition

Metric modulation is a change of tempo by reinterpreting a note value that two different tempos share. I'll show you how to calculate it and give some tips on how to notate and rehearse it.

Published August 7, 2026 · Updated August 21, 2026

What metric modulation is

According to Wikipedia,

In music, metric modulation is a change in pulse rate (tempo) and/or pulse grouping (subdivision) that is derived from a note value or grouping heard before the change.

For example, a pulse that was a dotted eighth before the tempo change might become a quarter note after it, but its length in milliseconds would be identical on both sides of the bar line. The new tempo is a consequence of the old tempo and the pivot note's value in each.

The term was first described by the critic Richard Franko Goldman, who used it in a 1951 essay on Elliott Carter's music. Carter preferred "tempo modulation": the meter may stay the same while the tempo changes.

The trick itself is older than the terms. Composers, such as Beethoven and Brahms, wrote proportional tempo relationships in their music years before Carter. His use of metric modulation wasn't subtle, however; he used it as a primary compositional technique.

Calcophony uses both terms. The Tempo Modulation section shows tempos in a matrix. The Metric Modulation section finds a path between a source and destination tempo using beam-limited breadth-first search, which this article does not cover.

The math

The pivot note has the same duration on both sides of the modulation, so how do we calculate the new tempo from the old one? The pivot equation follows from that shared duration.

Let p be the pivot's value counted in old beats, and q its value counted in new beats. A beat at the old tempo lasts 60 / oldBPM seconds, so the pivot lasts p x 60 / oldBPM. The same note at the new tempo lasts q x 60 / newBPM. Those are the same duration, so:

The calculation expresses p and q as fractions of a beat. A dot multiplies a note value by 3/2. A second dot multiplies it by 7/4. Halving or doubling the note value halves or doubles the number. An n-tuplet divides the space into n parts where a power of two would normally go, so a quarter-note triplet is 2/3 of a quarter and a quintuplet quarter is 4/5 of one.

Example: 120 BPM, dotted eighth becomes quarter

Take a passage at 120 BPM with a quarter-note beat, and modulate so that the old dotted eighth becomes the new quarter note. Work it in four steps.

1. Find the old beat in milliseconds. At 120 BPM a quarter note lasts 60000 / 120 = 500 ms.
2. Express the pivot in old beats. A dotted eighth is an eighth plus half an eighth, so it is 3/2 x 1/2 = 3/4 of a quarter. That is p = 3/4, or 0.75 x 500 = 375 ms.
3. Express the pivot in new beats. It becomes the new quarter note, which is the new beat, so q = 1.
4. Apply the equation. newBPM = 120 x (1 / (3/4)) = 120 x 4/3 = 160.

Verify the result by reversing it: at 160 BPM a quarter note lasts 60000 / 160 = 375 ms. That is the same 375 ms the dotted eighth occupied at 120. The note's duration stayed constant while its notation changed.

The multiplier 4/3 is the pivot's ratio turned upside down. That is a useful shortcut once you trust it: if the pivot is worth 3/4 of the old beat and becomes the whole new beat, the tempo goes up by 4/3. If it is worth 3/2 of the old beat and becomes the new beat, the tempo drops by 2/3.

A reference table of common pivots

These are the pivots that come up most often, all calculated from a starting tempo of 120 BPM with a quarter-note beat. The pivot duration remains identical before and after the change in every row.

The reachable tempos are sparse. From 120 BPM, the simple pivots land on 60, 80, 90, 96, 150, 160, 180, and 240, with nothing in between. The relationship is also symmetric: every modulation has an inverse, and the inverse of 4/3 is 3/4. If you can get from 120 to 160, you can get back by flipping the values around.

Multi-step modulations

The sparseness is the problem. Composers rarely want to land on 160; they want to land on the tempo the next section needs, and that tempo is usually not reachable in one pivot. The standard answer is to chain modulations, multiplying the ratios as you go. Two steps of 2/3 and 5/4 compose to 2/3 x 5/4 = 5/6, and the order does not matter to the destination, only to what the players have to execute on the way.

Take 120 BPM to 100 BPM, which needs a ratio of exactly 100/120 = 5/6. No single simple pivot produces that ratio. The reason is its factor structure:

Plain note values contribute factors of 2. Dots contribute factors of 3 (and 7, for a double dot). An n-tuplet contributes a factor of n. Those are the only prime factors a pivot can introduce. So a tempo ratio containing a factor of 5 is reachable only through a quintuplet, and no number of extra steps will remove that requirement.

The 5 in 5/6 is therefore non-negotiable. What you get to choose is where it sits and how exposed it is. Consider these routes from 120 to 100:

• One step, exact, hard. Re-read the quintuplet sixteenth (five per beat) as a sextuplet sixteenth (six per beat). At 120 BPM the quintuplet sixteenth lasts 500 / 5 = 100 ms; at 100 BPM the sextuplet sixteenth lasts 600 / 6 = 100 ms. The route is exact and demanding in rehearsal.
• Two steps, exact, playable. Dotted quarter becomes quarter, taking 120 to 80. Then quintuplet quarter becomes quarter, taking 80 to 100. Both pivots are in the table above, and the quintuplet sits at the beat rather than four levels down.
• One step, inexact, easiest. If quintuplets are off the table for this ensemble, re-read the triplet eighth as a septuplet eighth. That lands on 102.86 BPM, which is 2.86 BPM sharp of the target. Whether that matters is a musical judgment, not a mathematical one.

Manual calculation becomes difficult as the number of steps grows. Two steps require multiplying fractions while preserving both denominators; three steps require enumerating pivot pairs at every stage and checking each intermediate tempo against what the players can hold. Rounding an intermediate tempo to a whole number changes the ratio and can produce a final tempo you did not choose.

Notating and rehearsing it

Write the equation above the barline in the conventional form: the old note value, an equals sign, the new note value. Put the resulting metronome mark next to it in parentheses. Performers need both. The equation tells them how to get there; the number tells them whether they arrived.

Ideally, the pivot should be audible before the modulation, not introduced at it. If the new tempo's quarter note is going to be the old dotted eighth, that dotted-eighth pulse should already be perceptible in some clear manner beforehand. Though an artistic case can be made for a stealthier approach, it can make it harder to land on the correct tempo.

It may also help to rehearse the chain one step at a time. Players can usually hold a single pivot on the first attempt; a three-step chain taken at speed is a different task, and hearing each landing tempo in isolation first makes the whole sequence stick.

Letting a calculator do the arithmetic

The arithmetic is tedious and error-prone. That work can discourage composers from trying the fourth or fifth option when the second one nearly worked.

Calcophony's Tempo Modulation module is the reference table above, generalised: enter a source tempo and it renders the full ratio matrix across every note value, dotted value, and tuplet, showing what each pivot pair yields. Enter a target BPM and it highlights the cells that hit it. Tap any cell and it plays the modulation so you can hear whether the pivot is useful.

The Metric Modulation module handles the multi-step case. Give it a start tempo, a target, a step limit, a tuplet ceiling, and a BPM tolerance, and it searches pivot chains breadth-first and ranks what it finds, reporting each path's final deviation, its cumulative tuplet complexity, and the estimated duration of the transition. These values let you compare routes to 100 BPM directly. Both modules work in exact rational arithmetic rather than decimals, so intermediate steps never drift.

Related modules

← All articles