The rule of thumb and what it hides
One millisecond per foot is the figure everybody carries around, and it is close enough to set a delay on a small system. At 68 degrees Fahrenheit the real number is 0.888 milliseconds per foot, so the rule of thumb overshoots by about 11 percent. Over 30 feet that is 3 milliseconds and nobody notices. Over 300 feet it is 33 milliseconds, which is not a rounding error, it is an audible echo.
The temperature term is where it comes from. The speed of sound in air rises by roughly 0.6 metres per second for every degree Celsius, so cold air is slow air. A show that soundchecks at 85 degrees and plays at 60 has had the speed of sound drop by about 15 feet per second, which over a 200 foot throw shifts the arrival by around 2.5 milliseconds. On its own that is subtle. Combined with a delay ring that was set for the warm air, it smears the transition rather than tightening it.
Humidity affects the speed too, and by so little compared with temperature that it is not worth a field. What humidity does affect strongly is high frequency absorption over distance, which is a different problem with a different solution.
Why the offset exists at all
If a delay speaker is timed to arrive at exactly the same moment as the main system, the listener standing under it gets two equally-early arrivals from two different directions and the image sits nowhere in particular. Add a few milliseconds so the main system wins the race, and the ear localises towards the stage even though the nearby speaker is louder. That precedence behaviour is the whole reason delay rings work rather than just adding confusion.
How much offset is a matter of taste and of how far apart the two systems are in level. Five milliseconds is a light touch. Fifteen is common on large systems. Beyond about twenty five to thirty milliseconds, with a big level difference, it stops reading as one source and starts reading as an echo, and where exactly that happens depends on the material — speech gives it away long before music does.
The offset is also the reason a second delay ring is not simply the first ring plus more time. Each ring is timed from the main system, not from the ring in front of it, so both carry the same offset and neither accumulates.
Measure source to source, not along the ground
The distance that matters is between the acoustic sources, and on a real site those are rarely both at head height. A main hang 25 feet in the air and a delay speaker on a 12 foot stick 180 feet away are 180.5 feet apart, which is close enough to ignore. The same main hang and a delay ring only 40 feet away are 42 feet apart, and that 2 feet is nearly 2 milliseconds — proportionally a far bigger error, because the near ring has less flight time to hide it in.
The other measurement mistake is timing from the wrong reference. If the delay is being set from the console rather than from the amplifier output, whatever latency the processing chain adds is already in the path, and it has to come out of the number rather than being stacked on top of it. Anything with a network audio transport, a digital crossover or a wireless link has latency, and it is usually published.
What this cannot do
It cannot align a subwoofer to a main box. It gives the time that corresponds to a physical offset, which is a starting point, but the crossover filters between the two boxes shift phase around the crossover frequency and the acoustic centre of a horn-loaded box is not where the front grille is. The gap between the tape measure answer and the measured answer is routinely two or three milliseconds, and the way that gets settled is with a measurement system and a transfer function, not with arithmetic.
It also has nothing to say about coverage. A delay speaker fires into a zone, and where that zone starts and stops is a question about the box pattern and the aim. The PA speaker SPL calculator covers the level side — how loud the ring has to be to be worth adding, and how much of the main system it is replacing at that distance. For the cable getting the signal out there, the AV cable run calculator covers run lengths and where passive copper stops working.
Questions people ask
How many milliseconds per foot does sound travel?
About 0.89 milliseconds per foot at 68 degrees Fahrenheit, which is where the familiar one millisecond per foot rule comes from — it is roughly 11 percent long. Put the other way round, sound covers about 1.13 feet per millisecond. The figure changes with temperature: colder air is slower, so a given distance takes more milliseconds on a cold night than it did at the warm afternoon soundcheck.
How do I set the delay on a delay speaker?
Measure the straight-line distance from the main speakers to the delay speaker, source to source rather than along the ground, convert it to milliseconds at the air temperature you will actually play in, then add a small offset so the main system arrives first. Five to fifteen milliseconds is the usual range for that offset. Subtract any latency the signal path adds before the delay speaker, because that is time already spent.
Why add extra delay instead of matching the arrival exactly?
Because the ear locates a sound by whichever version reaches it first, even when a later arrival is louder. If the delay speaker and the main system arrive together, the listener standing underneath gets no directional cue and the image collapses onto the nearest box. Letting the main system arrive a few milliseconds early keeps the apparent source on stage. Overdo it and the two arrivals separate far enough to be heard as an echo, which is the failure the offset is meant to avoid.
Does temperature really change the delay time?
Yes, and on long throws it is audible. The speed of sound rises with temperature, so a 40 degree Fahrenheit swing changes the flight time over 200 feet by roughly five milliseconds. A system aligned at a hot afternoon soundcheck is misaligned by that much at a cool night show. Short throws hide it, long ones do not, which is why outdoor systems with distant delay rings get rechecked once the air settles rather than trusting the afternoon numbers.
Can I use this to time-align a subwoofer?
Only as a starting point. The physical offset between a sub and a main box converts to a time, and this page will give you that time, but the acoustic centres of the two boxes are not at their front grilles and the crossover filters shift phase around the crossover region. The result is that the geometric answer and the measured answer commonly differ by a couple of milliseconds, and the measurement wins. Use the tape measure figure to get close, then measure.