Skip to content

Why tiny timing differences create human feel

A small, consistent offset makes a part lean back or push. Random humanize and large offsets do the opposite. At 90 BPM, a 20 ms lean is 12 percent of a 16th.

Updated · 7 min read Lesson 11 of 20 in Arrangement & Groove

In short

  1. Microtiming is a feel control: a small, consistent lean makes a part sit back or push, while large offsets lower the groove.
  2. At 90 BPM a 16th lasts 166.7 ms, so a 20 ms lean is 12 percent of a step, heard as together but no longer fused.
  3. Move one secondary part by the same amount on every hit and keep the kick and snare as the anchor.
In this lesson (8)

The beat that went flat

You record a drum part, select everything and it to 100 percent. The snare lands exactly on beats two and four. The hats sit exactly on the 16th-note lines. On screen it is perfect. On playback something has gone flat. The vocal floats on top of the beat instead of sitting in it, and the hats tick like a clock.

Quantizing removed a relationship between the parts. Before, the hats leaned slightly behind the kick, the same way on every hit, and that lean was part of the feel. After, every part shares the same instant, so nothing pushes and nothing sits back.

Why it matters

Feel is a lean, not a wobble

is the set of small offsets, from a few to a few tens of milliseconds, that place notes ahead of or behind the grid. Players use them on purpose. A drummer who sits behind the beat makes a groove feel relaxed. A percussionist who plays slightly ahead pushes it forward.

That does not make the grid the enemy. In listening tests, fully quantized versions of real bass and drum grooves were rated as grooving as much as the original performances, and exaggerated offsets lowered the ratings (Senn et al., 2016). Other studies found the same drop as offsets grew (Davies et al., 2013; Frühauf, Kopiez and Platz, 2013). Microtiming does not add groove by itself. It is a control for feel: it decides whether a part leans back or pushes, and it only works when it is small and consistent.

Figure 1. Two beats at 90 BPM, where one 16th lasts 166.7 ms. The consistent lean moves every hat 20 ms late, so the hats sit behind the kick as one part. Random humanize scatters them up to 20 ms either side with no pattern.

Listen

From quantized to played

Add timing drift to the hats and snare. A few milliseconds of consistent drift sounds like a drummer. Past a point it sounds loose.

Loading demo…
100%

Start with your speakers or headphones low.

The science

How big an offset really is

The offset of a note is its measured arrival minus its grid position:

toffset=tactual−tgridt_{\text{offset}} = t_{\text{actual}} - t_{\text{grid}}

A positive offset is late and a negative one early. To judge its size, compare it with the length of one grid step. A 16th note lasts:

t16=60 0004×BPM mst_{16} = \frac{60\,000}{4 \times \text{BPM}} \ \text{ms}

At 90 that is 166.7 ms, so a 20 ms lean is 12 percent of a step. That is far too small to hear as a wrong note, but large enough to change how two parts sit together. Rasch (1978) found that notes starting up to about 30 ms apart still sound as if they start together, yet the ear uses that head start to hear them as separate sounds (Moore, 2012). A hat 20 ms behind the kick still lands with it, but the two stop fusing into one hit.

Figure 2. A 20 ms lean against the length of a 16th note. Onsets up to about 30 ms apart still sound as if they start together, so the lean reads as feel rather than as a wrong note. A 16th is five to eight times longer than the lean, depending on tempo.

The pattern of the offsets matters as much as their size. Hennig and colleagues (2011) measured human rhythmic performance and found that timing errors are not independent from hit to hit. They drift in slow waves, so each error is related to the ones before it, even many beats back. Listeners preferred a beat humanized with that kind of related drift over one humanized with plain random offsets, the kind most humanize functions use.

Try it

Consistent lean against random drift

0 of 7 steps done

The +20 ms hats sit back and the -15 ms hats push, and both sound deliberate. The random version covers the same range and sounds loose, because no two hits agree on where the beat is.

Common mistake

Random humanize as a shortcut

To make a track feel played, you select every note and run a random humanize. Random offsets jump independently from one hit to the next: 12 ms late, then 10 ms early, then on the grid. A real player does not do that. Their timing drifts, and when they lean, they lean the same way for a reason, such as a snare that sits a little late on every while the kick stays with the bass.

The second mistake is assuming that more offset means more feel. Past the amount a good player would use, offsets stop sounding like a lean and start sounding like a mistake, and listeners rate the groove lower.

Takeaway

Lock the anchor, lean one part

Keep the kick and snare tight to the grid unless you have a reason to move them. They are the anchor the listener locks to. Then choose one secondary part, such as hats, shaker or a percussion loop, and move all of it by the same small amount. Set it by ear, not by eye, and check it from across the room. If the part sounds like one player with an attitude, keep it. If it sounds like two players disagreeing, pull it back toward the grid.

3 questions

Check yourself

  1. At 90 BPM, how long does one 16th note last?
  2. Why does random humanize often sound sloppy rather than played?
  3. A 20 ms lean on the hats gave your 90 BPM beat a relaxed feel, so you try 60 ms to make it groove harder. What do the listening tests on bigger offsets predict?

Sources

Show all 6
  • Davies, M., Madison, G., Silva, P., & Gouyon, F. (2013). The effect of microtiming deviations on the perception of groove in short rhythms. Music Perception, 30(5), 497-510.
  • Frühauf, J., Kopiez, R., & Platz, F. (2013). Music on the timing grid: The influence of microtiming on the perceived groove quality of a simple drum pattern performance. Musicae Scientiae, 17(2), 246-260.
  • Hennig, H., Fleischmann, R., Fredebohm, A., Hagmayer, Y., Nagler, J., Witt, A., Theis, F. J., & Geisel, T. (2011). The nature and perception of fluctuations in human musical rhythms. PLoS ONE, 6(10), e26457.
  • Moore, B. C. J. (2012). An Introduction to the Psychology of Hearing (6th ed.). Emerald.
  • Rasch, R. A. (1978). The perception of simultaneous notes such as in polyphonic music. Acustica, 40, 21-33.
  • Senn, O., Kilchenmann, L., von Georgi, R., & Bullerjahn, C. (2016). The effect of expert performance microtiming on listeners' experience of groove in swing or funk music. Frontiers in Psychology, 7, 1487.

Next lesson · 12 of 20

Swing explained without mystical languageSwing is one timing rule: every second note lands late. Here is the maths of the ratio, why tempo changes it, and how to stop parts from flamming.

Previous: End the song before the hook wears out

Portrait of Virzy Guns

Written by Virzy Guns

Songwriter and producer, founder of Virzy Guns Production. Now building HealingWave.