Why noise cancelling silences an engine but not a conversation

This looks at active noise cancellation from the practical end — what holds up once conditions stop being ideal.

Cancelling a sound means producing its mirror image

A microphone measures the pressure wave arriving near your ear, and the processor generates a wave of matching size and opposite sign. Where the two overlap, the pressure fluctuations subtract, and the ear receives something much closer to the silence between them. The cancellation is only ever as good as the copy, so any error in amplitude or phase leaves a residue you can still hear.

A copy half a cycle out of alignment does not cancel at all but adds instead, which is why poor implementations sound worse than none. This is arithmetic aimed at a moving target, and the target shifts every time you turn your head or the earpiece seal changes.

The microphones sit in two places for two reasons

A feedforward microphone on the outside of the earpiece hears noise before it reaches your ear, which gives the processor a head start. A feedback microphone inside the cup hears what actually arrived, which lets the system measure and correct its own errors afterwards. Feedforward alone is fooled by wind and by any leak in the seal, because it never checks the result of its own work.

At the protocol level, feedback alone cannot act early enough at higher frequencies, because by the time it hears a sound that sound has already been heard. Most designs combine both, which is why an earpiece has grilles facing outward and a small port pointing towards your ear canal.

Physics gives the processor a deadline

Sound covers roughly a third of a metre every millisecond, so the gap between the outer microphone and your eardrum is a few hundred microseconds. Everything has to happen inside that window: conversion to digital, filtering, generation of the opposing wave, and conversion back into sound. Low frequencies are forgiving because a single cycle lasts many milliseconds, which leaves room for error in both timing and processing.

Mechanically, a high frequency cycle can be shorter than the processing delay itself, at which point the correction arrives for a wave already gone. That deadline, rather than raw processing power, is why engine rumble disappears and the hiss of a nearby keyboard stubbornly does not.

Above a few kilohertz the seal does the work

Passive isolation is mass and sealing, a physical barrier that blocks sound before any microphone or processor becomes involved at all. Short wavelengths are blocked well by a snug ear tip, which is why in-ear designs cut treble noise better than loose pads do.

A leak the width of a hair barely matters at high frequencies but destroys low frequency cancellation, because bass escapes through any gap. This is why changing ear tips can transform cancellation on hardware that has not changed in any other respect whatsoever.

Manufacturers test the seal by playing a short tone and measuring what returns, which is exactly what a fit test in a companion app performs.

The strange pressure sensation has a cause

Many people report a feeling of pressure with cancellation switched on, similar to descending in an aircraft but without any real pressure change. The ear expects low frequency sound and low frequency motion to arrive together, and cancellation removes the sound while the motion remains.

The eardrum is also being held still against ambient fluctuations, which some people perceive as a steady loading rather than as quiet. Transparency modes, which deliberately pipe outside sound back in, usually relieve the sensation because the mismatch that caused it disappears. Sensitivity varies enormously between individuals, and there is no good evidence that the sensation reflects any harm to the ear.

Implementations differ, and vendors are not obliged to document the differences.

What cancellation costs you

The microphones, converters and processing run continuously, so cancellation shortens runtime noticeably compared with the same earpiece playing passively. Wind is the hardest case, because turbulence across the outer microphone is noise the microphone created rather than noise arriving from outside.

Under load, systems detect wind and reduce or disable feedforward cancellation, which is why the effect seems to vanish outdoors on a rough day. Cancellation also alters the frequency response of whatever you are listening to, so the tuning must compensate whenever the mode changes. That compensation is why music can sound subtly different between cancellation and transparency on identical hardware playing an identical track.

The takeaway

Cancellation buys you the low rumble; the seal and nothing else buys you the high hiss.

The constraint is almost always physical, and marketing rarely mentions which one.