the spectrum analyzer is making you deaf

you’re hunting a harsh frequency. you pull up the analyzer, lean in, and sweep a narrow band looking for the spike that matches the thing you think you heard. twenty seconds later you’ve found a peak at 3.1 kHz and you’re certain that’s the one. here’s the uncomfortable part: for most of those twenty seconds you weren’t really hearing the track. you were reading.

the frame

staring at a meter doesn’t just distract you from listening. it measurably turns your hearing down. vision and hearing draw from one pool of attention, and when the eyes work hard the ears go quiet: not as a figure of speech, as a suppressed signal in your auditory cortex.

this has nothing to do with discipline, or with focusing harder. it’s built into how attention is shared across your senses.

the experiment that should bother you

Nadine Lavie’s lab ran the clean version of this. in Macdonald and Lavie (2011), people did a visual task that was either easy or hard, and at one moment a plain, perfectly audible tone played at the same time. under the hard visual task, 79 percent of people didn’t notice the tone at all. not “heard it quieter.” didn’t hear it. they named the effect inattentional deafness.

read that number again. the tone wasn’t faint. the only thing that changed between hearing it and missing it was how busy the eyes were.

a few years later, Molloy, Griffiths, Chait and Lavie (2015) went looking for it inside the brain with magnetoencephalography. under high visual load, the early auditory response, the one that fires about a tenth of a second after a sound arrives, was reduced, and the later “i noticed that” response was suppressed on top of it. the same people got measurably worse at detecting the tone. the eyes and the ears were drawing from one account, and the authors described it as a push and pull: load the vision, starve the hearing. it has held up since, in a quieter lab (Raveh and Lavie, 2015) and in a flight simulator (Causse and colleagues, 2016), which is a more alarming place to stop hearing things than a mix.

now put a screen full of graphs in front of it

the spectrum analyzer. the gain-reduction meter. the correlation read-out. the moving waveform. every one of them is a hard visual task, and you run them at the exact moment you’re trying to make a careful listening decision. the tool you opened to help you hear is spending the attention you needed in order to hear.

(this is the part i find unsettling, because i build the displays.)

the harsh resonance you’re squinting to locate is, for those seconds, the precise thing your hearing has turned down so it can read the graph. you’re not hearing the mix and confirming it on the meter. you’re reading the meter and assuming you heard the mix.

the fix is almost stupidly simple

a meter doesn’t lie, and it’s a wonderful way to confirm a decision. the mistake is using it to make the decision. decide with your ears, then look to check what you already heard. the moment that order flips, the eyes start writing the verdict and the ears just co-sign it.

so close them. turn the screen off, or look away, or bypass and listen with nothing to read. make the harshness call blind. mark roughly where it sits and how much it bothers you. then open your eyes and let the analyzer tell you whether you were right. you’ll be right more often than the graph-first version of you would believe, and the handful of times you’re wrong are the times worth knowing about.

do the same with your A/B: switch blind, with your hand off the mouse and your eyes shut, and decide which one you prefer before you let yourself see which one it was. the version of you that can see the labels isn’t a neutral judge.

the same shape outside the studio

you already know this one. someone tells you something that matters while you’re reading your phone, and the words go in and none of them stay, and you have to ask them to say it again. the cost of looking was not-hearing. it’s the same single pool and the same push and pull whether the screen in front of you is a phone or a spectrum analyzer. attention isn’t free, and vision quietly bills first.

the tools

this is most of the reason i keep the KERN plugins visually quiet, and why CHECK hands you one steady verdict and one number instead of a wall of flickering detail to fall into: the less there is to read, the more of your attention stays where the decision actually gets made, which is in your ears. when you want to hear one thing on its own, the DELTA listen on SMOOTH plays you only what it’s removing, so you can judge it with your eyes shut and the meter closed.

i’m not telling you to throw the analyzers away. i’m telling you they belong in second place. look to confirm. don’t hand them the call.

from the studio

tell me the thing you only heard once you closed your eyes or turned the screen off: the resonance that was gone from the graph but not from the room, the part you muted by accident and only then realised was the actual problem. i collect these, and i read every reply.

jonas