Working with audio engineers  is a ton of fun. Davin Huston and I were just discussing the Fletcher-Munson curve, which describes how (objectively) loud certain frequencies need to be in order for a normal-hearing person to perceive them as being at the same volume. The normal hearing human ear is more sensitive to some frequencies than others.

Turns out that 3kHz is one of those frequencies. I'd never heard of this before. It's something that all (normal) hearing adults are more sensitive to than other frequencies -- soft 3kHz noises sound particularly loud and annoying to us. Only that frequency.

"It's the frequency that babies cry at, " Davin (who has a newborn) said.

I blinked. "And everyone with normal hearing has this bump, this sensitivity to 3kHz."


"Do all babies cry at this frequency?"

Yeah, said Davin. Doesn't matter if they can hear or not. It's a matter of the air being forced from tiny lungs through a tiny vocal tract. Babies are tiny didgeridoos.

This explains a lot. I know screaming babies actually make sounds -- they don't just lie there with their mouths open, which is what it looks like to me. (3kHz is so hilariously outside my hearing range that we've never even tried to amplify it -- my cochlea is so completely damaged there that you'd just be throwing data into a void.)

But I've never understood why screaming babies seemed to be so particularly annoying, gauged by the frequency and intensity of complaints I'd get from hearing travelling companions when we passed by a vocalizing infant. I used to wonder if it was just because it was a kid, and some sort of psychological "humans, take care of our species!" thing kicked in.Was a screaming baby more annoying than, say, a vuvuzela at the same amplitude?

Turns out the answer is yes. A screaming baby is physiologically more annoying than a vuvuzela played at the same amplitude, because it's something our species came up with in order to ensure we would take care of screaming babies of our species. We want them to shut up.

It also now makes more sense to me how it's technologically possible to have flashing alerts for crying babies (for Deaf parents). When I encountered the idea of "flashing alerts for crying babies" last summer (Lynn and Sharon were brainstorming on a hypothetical future-Mel-as-a-mom house -- long story), my first thought was "oh my gosh, that must be a really complicated signal processing problem. All those babies, all those variants of voices, all those variants on crying -- how can you train the dang thing to hear your baby?"

But nope. All they need to do is filter for a 3kHz sound. Easy.

I'm not sure why this never occurred to me before -- that baby-output might be simple. I guess I have a default assumption that humans make really complicated sounds. Which is usually true. But sometimes... not true.

So! Babies are vuvuzelas. Who knew?