The whisper the faucet swallows
Turn on the kitchen tap and try to keep talking. Your own voice doesn't get quieter—the sound waves leaving your mouth are exactly as strong as they were a moment ago—but somehow the words go missing under the rush of water. Turn the tap off and they snap back into the room. Nothing changed about the speech. Something changed about your ability to hear it.
This is auditory masking: the everyday fact that one sound can render another inaudible, even when the quieter sound is still physically present and reaching your ear. It is not a failure of attention and not a trick of the mind wandering. It happens in the machinery of the inner ear itself, and once you understand where, a lot of ordinary listening experiences stop feeling random.
A piano with overlapping strings
To see why masking happens, picture what a sound actually does inside your head. Deep in the cochlea sits the basilar membrane, a coiled ribbon of tissue that acts like a frequency analyzer. It is stiff and narrow at one end and floppy and wide at the other, so different pitches make different spots along it vibrate most. High frequencies peak near the base, close to where sound enters; low frequencies travel all the way to the far apex. Sitting along the membrane are the hair cells that convert that motion into nerve signals. In effect, your ear lays sound out on a map, pitch by pitch.
But the map is not made of sharp pixels. When a tone arrives, it doesn't excite one lonely point—it sets up a traveling wave that humps up over a small region and tapers off around it. Hearing scientists call these regions critical bands: the ear analyzes sound not in infinitely fine slices but in overlapping neighborhoods of frequency. Two sounds that fall inside the same critical band aren't handled by separate detectors. They compete for the same patch of membrane and the same hair cells.
That competition is the whole story. When a loud sound drives a region of the basilar membrane into large vibration, a softer sound landing in the same neighborhood adds almost nothing detectable on top. The hair cells are already swinging; the small extra push gets lost. The quiet sound is still there in the air and still moving your eardrum. It simply cannot raise its voice above what the loud sound is already doing in that spot of the cochlea. The faucet and your speech share enough frequency territory that the water wins.
The masking is lopsided
Here is the detail that surprises people: masking is not symmetric across pitch. A low, loud sound is very good at hiding higher, softer sounds—but a high sound is comparatively bad at hiding lower ones. Audiologists call this the upward spread of masking.
The reason is baked into the traveling wave. A wave for a given frequency builds gradually as it moves inward, then drops off sharply just past its peak. So the "skirt" of a low-frequency wave extends forward into the territory where higher frequencies live, splashing energy across their region. Higher frequencies don't return the favor, because their waves have already died out before reaching the low-frequency zone at the apex. This is why a deep rumble—a truck engine, an air conditioner, a bassy room—can bury the crisp high-frequency detail of speech far more effectively than a hiss would bury the bass. The consonants that carry meaning live up in that vulnerable high region, right where low noise reaches to smother them.
Masking that reaches across time
Stranger still, masking isn't confined to the instant two sounds overlap. A loud sound can hide a soft one that comes slightly after it, and even, briefly, one that came just before it. Researchers call these forward masking and backward masking.
Forward masking is the more powerful of the two and the easier to intuit. After a loud sound stops, the auditory system doesn't reset instantly; the hair cells and neurons that just fired hard need a moment to recover their sensitivity. A faint sound arriving in that recovery window—up to roughly a couple hundred milliseconds afterward—can slip by unheard. Backward masking, where a sound is hidden by a louder one that follows it, works over a much shorter span and points to the fact that the brain integrates incoming sound over a small window rather than registering each instant cleanly. In both cases the ear is revealing that hearing takes time, and that time has edges where information falls through.
Your streaming music is built on this
Masking isn't just a quirk to notice—engineers have spent decades exploiting it. Every MP3, AAC file, and streaming audio codec you have ever listened to relies on it directly. These formats use a psychoacoustic model: a mathematical map of what the human ear can and cannot hear in the presence of louder sounds. During compression, the encoder calculates a masking threshold for each moment of music, then throws away the sonic detail sitting below that threshold—the parts your cochlea would have discarded anyway. That is how a song shrinks to a tenth of its original size while still sounding, to most ears, untouched. The file isn't a perfect copy. It's a copy stripped of everything masking already hides from you.
So the same mechanism that loses your voice under the tap is what lets an entire library fit on your phone. The ear's limitation became the engineer's budget.
What masking reveals about your own hearing
Understanding masking reframes a frustration many people carry. If you struggle to follow a conversation in a rumbling car, a humming office, or a restaurant with a low drone, the instinct is to blame focus or to assume your ears are failing outright. Often what you're meeting is masking doing exactly what it does to everyone—amplified, perhaps, by how your particular hearing is shaped.
Because here is the link that matters: masking and hearing loss push in the same direction. When high-frequency sensitivity fades—the most common pattern of age- and noise-related hearing change—the soft, high consonants are already faint. Add the upward spread of masking from any low background noise and those consonants disappear entirely. Two people can sit in the same noisy room; the one whose high-frequency thresholds have slipped loses far more of the conversation, because the sounds they most need are both the quietest to them and the easiest to mask. This is why "I can hear fine at home but not in restaurants" is such a common and real experience, not an excuse.
Which is also why paying attention to the shape of your hearing—not just whether you "pass" or "fail," but which frequencies are strong and which are quietly slipping—tells you something useful about how the world will sound in noise, long before it becomes obvious.
Where this connects
Audra is built around that idea of paying quiet attention. Its at-home pure-tone screening walks through the frequency range one pitch at a time, so you can see the actual contour of your hearing—where you're sharp, where the high edge may be softening—rather than a single pass-or-fail verdict. Tracked over time, that contour is exactly the information that predicts when background noise will start eating your conversations, and it does it all on your device, privately. It won't diagnose or treat anything, and it doesn't pretend to. It just helps you notice, early and honestly, the parts of your hearing that masking is best at hiding from you.
If you're curious what your own frequency map looks like, you can try a screening at audra.lumenlabs.works.