You have been shouted at by someone who loved you and had no idea they were doing it. Maybe in a car with the windows down. Maybe across a restaurant table, over the espresso machine, while they told you something tender at the volume of a fire drill. And you have done it back. You have leaned into a bar conversation and heard your own voice arrive louder than you meant it, a stranger's voice using your mouth.

Here is the uncomfortable part: you did not decide to do that. A reflex older than language decided for you, and it kept the decision from you until the moment the music cut out and you were left standing there, still yelling, in a room that had gone quiet.

The reflex that turns up your voice

In 1911, a French ear, nose, and throat physician named Étienne Lombard noticed something strange while testing patients he suspected of faking deafness. When he piped noise into their ears, their voices rose — automatically, without instruction, without awareness. Even patients who claimed they could hear nothing at all raised their voices when the noise came. Their ears were listening; only their stories weren't.

What Lombard found has been replicated for more than a century and now carries his name. In noise, speakers involuntarily increase vocal intensity. Across many studies, the trade is roughly half a decibel of extra voice for every decibel of added background noise — enough to keep you audible, never quite enough to fully restore the clarity you had in the quiet.

And it isn't just volume. Under noise, your voice climbs slightly in pitch. Your vowels stretch a little longer. The energy in your voice shifts upward in frequency, away from the low rumble that room noise drowns anyway and into the higher bands where consonants live. You articulate more crisply. Your speech becomes, without your permission, a better signal.

This is not a human quirk. Marmosets do it. Songbirds do it. Beluga whales raise their calls in shipping lanes. Infants who cannot yet form words cry louder into noise. Something this widespread across evolution is not a habit. It is plumbing.

Your voice is on a loop you can't see

To understand why the Lombard effect grabs you so completely, you have to accept a strange fact about talking: you have never heard your own voice the way anyone else does, and you have never spoken without listening.

Every time you speak, sound travels back to your own cochlea by two routes at once. Some leaves your mouth, travels through the air, and enters your ear canal. Much of it never leaves your head at all, vibrating up through the bone of your skull directly into the inner ear. Together these form what engineers call sidetone — the private, boomy, bass-heavy version of yourself only you will ever hear.

Your brain is not merely enjoying this signal. It is using it. Auditory feedback is a control loop for speech production, continuously comparing what you intended to say against what you appear to be saying, and correcting on the fly. This is why hearing a delayed recording of your own voice through headphones makes fluent speakers stutter within seconds. It is why singers close their eyes and cup a hand to one ear. The loop is running whether you consult it or not.

The Lombard effect is that loop reacting to interference. When background noise masks your sidetone — when you can no longer clearly hear yourself — the system does the only sensible thing a control loop can do. It pushes harder. And because the adjustment happens below the level of deliberate thought, you experience only the output. You do not experience the decision.

This is the part people find genuinely eerie once they notice it. You can be told to keep your voice steady in noise. You can concentrate on it. It leaks anyway. Suppressing the Lombard effect is possible with effort and training, but it is like suppressing a blink against an oncoming object: you're not overriding a choice, you're overriding a reflex that ran before the choice existed.

There is also a social layer. The effect tends to be stronger when you're actually speaking to someone than when you're reading aloud into an empty room. Your voice isn't just fighting noise. It's trying to reach a face.

What your volume is quietly reporting

Now flip the whole thing around, because this is where a century-old curiosity becomes something worth paying attention to in your own life.

If your voice is regulated by how well you hear yourself, then how loudly you speak is a readout — an imperfect one, but a real one — of what your ears are doing.

Audiologists have long noticed a pattern that follows directly from the sidetone loop. When the problem is in the inner ear or the hearing nerve, everything sounds faint, including your own voice, so the loop compensates and people tend to speak more loudly than they intend. But when something blocks sound from getting in — wax, fluid, a bad head cold, a plugged ear — outside sounds are muffled while your bone-conducted voice arrives unchanged and now dominates. Your own voice booms. The loop backs off. People with a blocked ear often speak too softly, and are baffled when told so.

You can feel the second one right now. Plug both ears with your fingers and say a sentence. That sudden bloom of bass is the occlusion effect, and you'll notice you instinctively spoke quieter. That instinct is the Lombard reflex running in reverse.

Which is why one of the earliest signs of a hearing change is not something you notice. It is something other people notice about you. The partner who says you've started talking to them from three rooms away in a shout. The colleague who winces on a video call. Hearing shifts erode slowly, symmetrically, and your brain rewrites its baseline as it goes. But your voice keeps reporting to the room, honestly, and it doesn't tell you what it said.

There's a second thing worth sitting with. If you find that restaurants have gotten harder — that you're straining, leaning in, shouting more than you used to, going home tired — that difficulty is not merely the room getting louder. Speech-in-noise is one of the first things to go, and it goes early, sometimes while quiet-room hearing still tests fine. The exhaustion you feel after a loud dinner is real work: your brain paying overtime to reconstruct sentences from fragments.

Your next moves

  • Ask the two people who'd tell you the truth. Ask your partner, roommate, or closest colleague a specific question — not "do I talk loud?" but "in the last month, have you noticed me speaking louder than the room needed?" Vague questions get polite answers.
  • Run the finger test tonight. Plug both ears, say a full sentence, then unplug and repeat it. Feel how strongly your voice volume tracks your sidetone. That felt sense makes the reflex catchable in the wild.
  • Practice one deliberate override this week. Next time you're in a loud room, pick a single sentence and consciously hold your voice at conversational level while moving physically closer and facing the person directly. Distance and line-of-sight buy you more intelligibility than volume ever will — and you'll feel exactly how hard the reflex pulls back.
  • Notice the after-dinner tiredness. If you leave loud social settings drained rather than energized, write it down with the date. Two or three entries over a month is data, not a mood.
  • Get a baseline while you still think you don't need one. A hearing threshold measured now is worth far more than one measured after you're worried, because change is the signal and you cannot measure change without a starting point.

The instrument you can't read from inside

The Lombard effect is a small, elegant proof of something larger and slightly unsettling: your hearing regulates your behavior long before it enters your awareness. Your voice already knows. Your ears already knew. You are the last to be informed.

That's the gap Audra was built for — giving you a way to actually look at what your ears are doing rather than inferring it from the way people flinch at dinner. It runs a pure-tone screening on your phone, entirely on-device, so you can see your thresholds across frequencies and watch them over time instead of guessing; and for anyone whose ears also ring in the quiet, it builds personalized notched-noise sound enrichment tuned to your own profile. The screening is free, and it takes about as long as the dinner you'll go to tonight. If you'd like to know what your voice has been telling everyone else, you can start at audra.lumenlabs.works.