Field notes on things that run themselves
A Room Finds Its Own Voice
Every performer knows the half-second before it happens: a rising, rounding pitch climbing out of nowhere, swelling in under a second into a squeal loud enough to make an entire room wince. Nothing in the room just broke. The same microphone, amplifier, and speakers that were carrying a voice cleanly a moment before are still working exactly as built — they have simply found a much easier job than the one they were given, and started doing that instead.
The proper name is audio feedback, or the Larsen effect, and the loop hiding inside it is a genuinely closed circuit: a microphone, an amplifier, a loudspeaker, the air of the room, and back into the microphone again, around and around. The word feedback names exactly this — a system’s own output, fed back in as its own input, with nothing outside the loop required to keep it going once it starts.
As long as what comes back into the microphone on each pass is a little quieter than what just left the speaker, the loop stays stable — a voice enters once, gets amplified, and fades into the room like any ordinary sound. The instant that stops being true at any one frequency, that frequency stops behaving like an ordinary sound at all. It reinforces itself on every pass through the loop, doubling and redoubling many times a second, until within a fraction of a second it is the only thing the system is carrying. Electrical engineers have a name for the exact tipping point, the Barkhausen criterion — the same rule used to design every electronic oscillator ever built on purpose. Here, nobody designed it; the gear simply found a different, far more efficient way to be stable.
The counterintuitive part is that none of this is really about how loud the room is. It is about gain — how much amplification sits between what the microphone hears and what the speaker plays back. A powerful singer working close to the mic can run at very little gain and stay safe however loud the show gets, since the microphone barely needs boosting. A quiet performer working farther back needs the gain turned up just to be heard at all, and it is that turned-up gain, not the volume filling the room, that pushes the whole loop toward its threshold. The fader marked quietest is often the one standing closest to the edge.
Why does the loop lock onto one specific note instead of howling across every frequency at once? Because its willingness to reinforce a signal is never flat — it has peaks and valleys shaped by the microphone capsule’s own resonance, the loudspeaker’s uneven response off to the sides, any boost already dialed into the mixing desk, and the room itself. Sound bouncing between two hard, parallel surfaces sets up genuine standing waves of its own — the same physics that makes one particular note ring especially rich in a tiled shower or a stairwell — and those room resonances stack on top of whatever the equipment already favors. Whichever peak crosses the threshold first wins, almost instantly drowning out every other frequency in the room. The pitch of a howl is rarely one tidy cause; it is closer to an involuntary confession — a map of every resonance the room and the rig were already holding, silently, all along.
Sound engineers have turned finding that map into a ritual, done on purpose before an audience ever arrives. Ringing out a room means pushing the gain up deliberately until one frequency just starts to howl, noting exactly which, pulling it back down a few decibels on a graphic equalizer, and repeating until the resonances hiding in that room and that gear have been found and tamed before a single real note is played. Layout does more of the work than any equalizer afterward: a cardioid microphone’s dead zone should face the monitor speaker beside it, and a curtain hung over a hard back wall breaks up the reflections that would otherwise hand feedback a second path home. Modern feedback suppressors automate the same hunt, listening for one frequency building on itself and notching it out within a fraction of a second — a machine trained to recognize its own kind, and interrupt it.
The effect is named for Søren Absalon Larsen, a Danish physicist who began his career studying philosophy and theology before turning to physics and electrical engineering, eventually becoming a professor at what is now the Technical University of Denmark. In a short 1911 paper on an acoustic alternating-current generator, Larsen formally described this same closed loop — decades before public-address systems, let alone rock concerts, existed to lose control of one. He was not documenting a nuisance to be eliminated. He had found a genuinely new way to produce a steady, self-sustaining tone out of nothing but a loop and a threshold, and written it down.
Every self-sustaining loop this publication has covered so far has earned its keep: a heartbeat, a coral reef, a candle flame, a voice singing an actual word. This is the identical trick — a signal reinforcing itself, kept alive by nothing but continuous energy pouring through the loop — running inside a system built for an entirely different job, the moment one threshold gets crossed by accident. It took musicians barely over a decade to stop treating that accident as only a failure. During a break in an October 1964 recording session, John Lennon leaned his guitar against an amplifier and the room filled with a rising note; George Martin recognized it at once, and the Beatles kept it as the opening two seconds of “I Feel Fine.” Within a few more years, guitarists were leaning into their monitors on purpose, chasing the exact loop described above — not fixing the feedback, but aiming it.
One loop I’m watching
Next: a greylag goose whose egg has rolled from the nest will retrieve it with an extremely stereotyped neck-crooking motion — and famously, if the egg is lifted away mid-retrieval, the goose completes the entire motion anyway, rolling nothing at all back to the nest. The clearest version yet of a program that runs to completion whether or not the outcome it was “for” is still there.
Tip: the ← and → arrow keys move between issues.
Get the next issue the moment it’s live — subscribe by email or follow by RSS.
New to The Standing Wave? Start here →