the eleven true things, and the one i had to cut

i had twelve of these in the video. “under thirty milliseconds two sounds fuse.” “the ear integrates over four hundred milliseconds.” “take a break every forty-five minutes.”

every one has a real experiment behind it, with real conditions attached. what reaches you is the number with the conditions stripped off, because the number travels and the conditions do not.

so it survives the trip and stops being true on the way.

what fell off

almost none of these rules are wrong. they are measurements, taken under conditions, and the conditions fell off somewhere between the paper and you. so you end up applying a true finding in the one situation it was never measured in.

i went and read the papers behind all of them. eleven held up. the twelfth is the interesting one, and i’ll come back to it.

the thirty millisecond one

take the one everybody knows: under thirty milliseconds two sounds fuse into one, over thirty you hear an echo.

then i read the actual experiment. it measured something else: continuous speech with a single echo, outdoors, and asked whether the echo was disturbing, not whether you could hear two sounds. at equal level the answer came out at sixty-eight milliseconds, not thirty.

then the echo was turned down three decibels and the answer moved to a hundred and eight. six decibels down, a hundred and seventy-five.

level moved the answer by more than a hundred milliseconds. so a rule made only of milliseconds was never going to hold, and “disturbing” was never the same question as “i hear two sounds.”

and it kept showing up like that. the number stayed, the setup disappeared.

the one that costs you time

here’s one you can test tonight.

play a low tone through one speaker, in your room, and walk a metre. don’t touch anything else. the level at your ears will move, and it can move a lot.

i went looking for the number everyone quotes for this. what’s published is a simulation comparing positions to an average. it isn’t a measurement of you leaning forward, so i’m not going to hand you a figure for it.

nothing about the sound changed. the room did, because you moved through the pressure pattern of one standing wave. your seat is part of your monitoring, and there’s no universal number for how much, because it depends on your room.

so before you reach for the low end, move your head. if the bass changes when you lean, you’re hearing the room as much as the mix, and the fix is where you sit and what is behind you.

the meter one

three sounds can hit the same peak to the decibel and not be equally loud. a peak meter cannot see that, because it’s not measuring the thing you’re judging.

what got measured, when someone actually measured it, was a range rather than a number: between a five millisecond sound and a two hundred millisecond one the difference came out somewhere between ten and nineteen decibels, and it moved with level.

even the loudness standard doesn’t pretend harder than that. it calls its own output an estimate.

use a loudness meter for loudness. keep the peak meter for clipping. it won’t tell you loudness.

the one i had to cut

i wanted to tell you what share of listening happens on something that is effectively mono at the bottom. it’s the number that would make the whole argument land.

i couldn’t find one.

phone sales don’t answer it, because owning a phone tells you nothing about whether the sound left through it, through headphones, through a car, or through a speaker with two drivers in it.

so i can show you that side content disappears when a path sums to mono. i can’t tell you how often you’re on that path.

i cut it on camera rather than just dropping it, because “nobody has published this” almost never gets said out loud. and it beats a number i would have made up.

that was the rule for the whole video: if i couldn’t defend the popular version, i said so on camera.

so what do i actually do

not much, honestly. i don’t have a trick for this one.

mostly i just ask the annoying question before i trust a number: true for what setup? what was measured, on whom, at what level, for how long. the answer is usually narrower than the rule.

and if i can’t find the setup, i try to stop there.

the video goes through all eleven, with the measurement each one came from. there are demonstrations throughout, so headphones help.

watch the video at the top.