Noise cancellation

Noise cancellation is a signal processing technology that reduces or removes unwanted sound coming in from around you. Earphones and headphones create a waveform that is the opposite of the noise and add it to cancel the noise. For the signal coming in through a microphone, a method that keeps only the voice is also used. The core aim is to help people hear a voice clearly.

This technology came into wide use as the annoyance of noise mixing into phone calls and video meetings grew. Small devices such as glasses and earphones cannot carry many microphones, so research on raising sound quality with software became active. More recently, a method in which AI tells apart the features of noise and of voice is being added.

A device takes in the surrounding sound with a microphone, turns it into a digital signal and analyzes it. The active method makes a waveform with the same size as the noise but the opposite phase and adds it, so that the two cancel each other out.

The AI method keeps only the voice parts, using a model trained on the patterns of noise and the features of the human voice. It may also judge the direction a sound comes from by using the difference in the time the sound reaches several microphones.

It is used in earphones, headphones, smart glasses and video meeting equipment to raise call quality. Even if a device has few microphones, software can make up for it, so it is especially useful in small products.

To apply it, the microphone input must be stable and the chip must be powerful enough to process the computation in real time. Battery use and delay also have to be considered together at the design stage.

It does not remove every sound. Sudden loud sounds and noise that is similar to a human voice are hard to tell apart perfectly. In real environments, performance can vary a lot with conditions.

A common misunderstanding is that noise cancellation completely blocks all sound. In practice it mainly reduces noise in certain frequency ranges, so the range of its effect is limited.

It is like picking out only your friend's voice in a noisy cafe where many people are talking. The chatter around you becomes blurry, and the voice in front of you sounds clearer.

There are many tasks that need to take in sound accurately, such as calls, lectures and voice commands, so it helps to know this when you weigh sound quality claims in product ads or articles. It is also good to remember that even the same words can come out differently depending on the number of microphones and the software.