What Is Noise Cancellation? Adding Sound to Produce Silence

Sound is a pressure wave, and when two waves meet, the pressures add at every point. If a second wave is identical to the first but inverted, meaning its compressions coincide with the first wave's rarefactions, the sum is zero and both waves vanish where they overlap. That is destructive interference and it is ordinary physics, demonstrable with two loudspeakers. Making it work usefully against real noise requires the cancelling wave to match the unwanted one in amplitude and to be precisely inverted in phase, at every frequency present, at the exact location of the listener's ear, and continuously as the noise changes. Each of those is a hard constraint. A phase error of half a cycle turns cancellation into reinforcement, making things worse, and since a cycle at a high frequency lasts tens of microseconds, the timing precision required rises with frequency until it becomes unachievable. That single fact explains most of what the technology can and cannot do.

A practical system has a small number of components and several possible arrangements:

• •Feedforward, with a microphone on the outside of the earcup sampling noise before it arrives, giving the processor time to compute a response but no information about what actually reaches the ear
• •Feedback, with a microphone inside the cup next to the driver measuring the residual error and correcting it, which is accurate about the ear but has less time to act
• •Hybrid systems using both, which is what most current products do
• •A processor computing the inverted signal, historically analogue and now almost universally digital with adaptive filters that adjust continuously
• •The driver, which must reproduce the cancelling signal and the music simultaneously without distortion
• •Passive isolation from the physical earcup or tip, which does the work at high frequencies where active cancellation cannot, so the two approaches are complementary rather than alternatives

Low frequencies are long waves, with a cycle lasting many milliseconds, so there is ample time to measure, compute and emit a correction, and the wavelength is large enough that the cancellation holds across a region rather than at a single point. Aircraft cabin roar, engine rumble and ventilation noise sit exactly in that range, which is why the technology transformed long-haul flying. High frequencies are short waves, so the timing tolerance shrinks, the processing delay becomes significant, and the zone of cancellation becomes smaller than the distance the ear moves when the head shifts slightly. Speech, crockery, keyboards and sudden noises contain substantial high-frequency content and change rapidly, which makes them very difficult to predict and cancel, and this is why an open-plan office remains audible through headphones that silence a jet engine. Sudden impulsive sounds are hardest of all, since an adaptive system needs a pattern to work from and a bang provides none before it has already arrived.

The same principle is applied in rooms and vehicles with more difficulty, because cancelling over a volume rather than at one pair of ears requires many microphones and speakers and generally produces zones of quiet alongside zones where the sound is made worse. Car manufacturers use it against engine and road noise, sometimes combining it with deliberate engine sound synthesis, which means the vehicle is simultaneously removing and adding sound. Ducts and exhausts are a favourable case because the sound travels in one dimension along a pipe, which makes prediction tractable, and industrial duct cancellation predates consumer headphones by decades. Aircraft propeller noise cancellation has been fitted to turboprop cabins. The unresolved consumer questions concern safety, since removing environmental sound while walking or cycling is dangerous and has prompted transparency modes that deliberately pass sound through, and health, where the evidence that cancellation itself causes discomfort or pressure sensations is largely anecdotal and the more substantiated concern is simply that quieter surroundings invite louder listening.

The system emits a second sound inverted in phase so the pressures sum to nothing, which requires matching amplitude and phase precisely at the ear and continuously. Long low-frequency waves give enough time to compute and cancel across a region, which is why engine rumble disappears. Short high-frequency waves leave microseconds of tolerance and a tiny quiet zone, so speech and clatter get through and passive isolation handles them instead.