The whole calculation is one division
A quarter note in milliseconds is 60000 divided by the tempo. There are 60 seconds in a minute and 1000 milliseconds in a second, so a minute is 60000 milliseconds, and a tempo of 120 beats per minute means each beat gets 60000 / 120 = 500 of them. Everything else is a multiple of that: a half note is two quarters, a whole note is four, an eighth is half, a sixteenth is a quarter. Dotting a note adds half its own length again, so it is the straight value times 1.5. A triplet fits three notes where two would go, so it is the straight value times two thirds.
At 120 BPM that gives a quarter of 500 ms, an eighth of 250, a dotted eighth of 375 and an eighth triplet of 166.67 ms. Those four numbers cover most of what a delay plugin gets set to, and if you learn only one of them, learn that the dotted eighth is three quarters of a beat.
The reason to have a number at all, rather than clicking the tempo-sync button, is that plenty of things do not sync. Hardware units, plugins on a bus with a different clock, tape emulations with drift, and every delay you have to dial in on a stage all take milliseconds. Sync also fails whenever the delay should sit slightly off the grid on purpose, which is a common trick and one you cannot do from a menu of note values.
Choosing a note value that does something
Delay note values are not interchangeable. Each one has a different relationship to the part it is sitting behind.
| Value | At 120 BPM | What it does |
|---|---|---|
| Quarter | 500 ms | Repeats on the beat. Reinforces the pulse; disappears into a busy arrangement. |
| Dotted eighth | 375 ms | Lands between the eighths. Creates a cross-rhythm from a straight part. |
| Eighth | 250 ms | Doubles an eighth-note part. Thickens rather than adds. |
| Eighth triplet | 166.67 ms | Pulls a straight part toward a shuffle. Fights a straight groove. |
| Sixteenth | 125 ms | Short enough to read as thickness and ambience rather than a repeat. |
| 32nd and shorter | 62.5 ms and down | Stops being rhythm. Below about 40 ms it becomes comb filtering and tone change. |
The last row is the one that catches people. Very short delays do not sound like delays; they sound like the source has changed character, because the repeat interferes with the original across the frequency range rather than arriving after it. Two mics on the same cabinet at different distances produce exactly this, which is why mic placement and short delay are the same problem wearing different clothes.
Pre-delay is a delay, so it belongs here
Reverb pre-delay is the gap between the dry sound and the start of the reverb tail. Setting it to a musical value keeps the reverb from blurring the attack of whatever fed it, which is why a vocal with 40 ms of pre-delay sits forward of one with none even when the reverb amount is identical. A sixteenth or a thirty-second is the usual starting territory. Anything past a quarter note stops reading as a room and starts reading as a slap.
Decay time is a taste decision and the numbers on this page are only a way of not stepping on the next downbeat. A decay of one bar means the tail is roughly gone by the time the next section arrives. A decay of two bars on a dense track will keep everything wet through chord changes, which sometimes is the point and usually is not. Setting the decay by ear against the arrangement beats setting it from a table, but the table stops you accidentally choosing four seconds on a track where a bar is 1.7.
Feedback, repeats and the tail you did not budget for
Feedback is the percentage of each repeat fed back into the delay line. Each pass multiplies the level by that fraction, so the decay is exponential and the number of audible repeats grows quickly as feedback rises. At 40 percent each repeat is about 8 dB down and the tail is effectively gone after seven or eight of them. At 70 percent each repeat is only 3.1 dB down and it takes around seventeen repeats to fall 60 dB, which at a dotted eighth is over six seconds of tail.
Long tails are not a problem in isolation and are a serious problem in a mix. They accumulate low mids, they smear the transition between sections, and they make automation necessary where none was planned. The tail figure this page gives is the practical check: if the tail runs longer than the space between chord changes, either shorten it, filter the repeats, or duck them against the dry signal.
Hearing damage is the real hazard in this work
Sustained loud monitoring causes permanent hearing loss. It is cumulative across every session you have ever done, it does not hurt while it is happening, and by the time you notice it there is nothing to be done about it. It is also bad for the work: loud playback flattens your perception of the frequency balance and of dynamics, so decisions made loud tend to fall apart at normal listening levels. Monitor quietly by default, check loud briefly and rarely, and take breaks. If you work in loud rooms or on stages, wear protection made for music rather than foam plugs that remove the top end.
Where this connects
If you need a click rather than a number, the metronome plays the tempo directly. For the pitch side of the same session, the note frequency calculator converts note names to hertz and works out cents between two pitches. Once tracking starts, the audio session size calculator tells you what the take folder will weigh, and the room those decisions are made in is covered by the acoustic treatment calculator and the speaker placement calculator.
Questions people ask
What is the formula for delay time from BPM?
Milliseconds per quarter note = 60000 / BPM. At 120 BPM that is 500 ms. Multiply by the note value relative to a quarter: an eighth is 0.5, a sixteenth 0.25, a half note 2, a whole note 4. Multiply by 1.5 for dotted and by two thirds for triplet. Every value in the table above is that one division followed by a multiplication, which is why you can work out any of them in your head once you know the quarter note.
Why does the dotted eighth sound so different from the eighth?
A dotted eighth is three sixteenths long, so its repeats land on the offbeat sixteenths rather than on the eighths. Feed it a straight eighth-note part and the repeats fill the spaces, producing a pattern that cycles against the original every three notes instead of doubling it. A straight eighth delay lands on top of the notes that are already there, so it adds level and thickness but no new rhythm. That is the entire difference and it is why the dotted eighth is the default guitar delay on a great many records.
Should I set pre-delay to a note value or by ear?
Start from a note value and finish by ear. A sixteenth or a thirty-second gives you a gap that is rhythmically consistent with the track, which stops the reverb onset drifting against the groove. From there it is a balance question: more pre-delay pushes the dry signal forward and makes the reverb read as a separate space, less pre-delay glues the two together. On a slow ballad a sixteenth may be 190 ms, which is long enough to hear as a distinct event, so shorten it. On fast material a sixteenth is 60 ms and reads as depth. The note value scales with the tempo, which is exactly what you want.
Does the delay time change if my time signature is not 4/4?
The note values do not change. A quarter note is 60000 / BPM regardless of meter, because a beat in beats-per-minute is a quarter note by convention in almost every DAW. What changes is the bar length, which is why this page asks what counts as a beat and how many go in a bar. In 6/8 counted in eighths at 120 BPM, an eighth is 250 ms and a bar is 1500 ms rather than 2000. Delay and pre-delay come off the note table; anything phrased in bars comes off the bar figure.
Why does a very short delay change the tone instead of repeating?
Below roughly 30 to 40 milliseconds your hearing stops separating the repeat from the original and fuses them into one event. The two copies then interfere frequency by frequency, cancelling where they are out of phase and reinforcing where they are in phase, producing a comb-shaped response. The result is a hollow or metallic tone change rather than an echo. This is the same mechanism behind flanging, behind two-mic phase problems on one source, and behind the odd tone you get standing near a reflective wall. If you want thickness without the colouration, use a delay long enough to be heard separately, or detune and modulate the short one so the notches move.