the top octave is not always there when it arrives

there’s a genre of video that asks you to find the age of your ears. a tone climbs, and wherever it vanishes is supposed to tell you something about you.

i made a film about where that idea came from. while i was building it i ran a check i hadn’t seen anyone run, and it wasn’t on the ear.

what i actually did

i took the finished file from my last video, downloaded what the platform hands out for it, and put the two side by side.

it doesn’t serve one audio stream. it serves more than one, and which one you get is negotiated at play time by whatever you happen to be watching on.

the file my laptop gets still has everything up to 20 kHz. the file a lot of phones get stops dead at 16.

same upload. same video. the same second of audio. at 17 kHz that tone is either sitting there untouched, or it’s 55 dB down.

55 dB down is gone.

and which one you get isn’t something you choose. that’s one video on one platform, so read it as an example rather than a rule. nobody publishes a table saying a codec cuts at a given frequency, and opus’s 20 kHz is a ceiling, not a promise.

so the test has a condition sitting in front of it

if the tone is above 16 kHz, you’re being asked two questions and handed one answer.

does your ear respond up there? and did the stream that reached your player still contain it? from the listening chair you can’t separate those, and the result gets read as a fact about your ears.

the ageing is real. it’s well measured and i’m not arguing with it. it’s the test that’s unreliable.

if you want to know what your ears actually do up there, you need a tone generator running locally, on hardware you control, at a level you set. anything that arrives through a codec has already been edited.

the part you can use

anything you put above about 16 kHz isn’t something you can count on a listener receiving. not their speakers, not their ears. the delivery may or may not carry it, and you don’t get a vote.

which changes what “air” is worth. if a mix only reads as open because of what’s up there, it’ll read open for some people and flat for others, and you’ll never hear the difference on your own system, because your own system plays the file you made.

so treat the top octave as a bonus rather than as structure. put the thing that carries the mix lower, where every delivery path keeps it. shelve for air if you like the sound of it, then bypass the shelf and check the mix still stands up without it. if it collapses, the air was doing structural work, and that’s the first thing the pipe takes.

same shape as the low-end problem, from the other end. down there, width gets summed away. up here, the band gets thrown away.

ten minutes, on something you’ve already released

you don’t need my tooling for this.

take a track you’ve put out. download it back from wherever it lives. put the original and the download side by side in an analyzer and look above 15 kHz.

if you can force a different player or a different device, do it twice. that’s where the two streams show up.

you won’t un-know it.

i went in expecting to learn something about ears. what i ended up measuring was the pipe.