How Noise-Cancelling Headphones Work降噪耳机是怎么消掉声音的

They erase a jet engine by adding more sound to it. Drag the sliders, press play, and watch (and hear) how that works.

它们消除飞机引擎噪声的办法，是往耳朵里再加一段声音。拖动滑块、按下播放，亲眼看、亲耳听它是怎么做到的。

Sound is air being pushed around声音就是被推来推去的空气

Sound is a pressure ripple. A vibrating object shoves the air molecules next to it, they shove their neighbours, and the pattern travels outward at about 343 metres per second. The molecules themselves hardly go anywhere: each one only jiggles back and forth around its home position.

声音是一种压力的涟漪。振动的物体推挤旁边的空气分子，这些分子再推挤它们的邻居，这个图案就以每秒约 343 米的速度向外传播。分子本身几乎哪儿也没去，每一个只是在原位附近来回晃动。

Your eardrum only feels that curve: the pressure pressing on it, moment by moment. So if we could keep the pressure at your ear perfectly flat, you would hear nothing, even in a roaring cabin.

你的耳膜只感受得到这条曲线，也就是每一刻压在它上面的压强。所以，只要能让耳朵处的压强保持完全平直，哪怕身处轰鸣的机舱，你也什么都听不到。

Two sounds can add up to silence两个声音可以加成寂静

When two pressure waves meet, the air simply adds them. Push plus push is a bigger push. But push plus an equal pull is nothing at all. Below, the orange wave is a noise, the blue wave is a copy of it that we can slide in time, and the thick curve is what actually reaches the ear.

两列压力波相遇时，空气只是把它们简单相加。推加推，是更大的推；推加上一个同样大小的拉，就什么都没有了。下图里，橙色波是噪声，蓝色波是它的一个副本，可以在时间上前后滑动，粗线是真正到达耳朵的声音。

That is the whole idea of active noise cancelling: make an anti-noise that is the exact mirror image of the incoming sound, and play it into your ear at the same moment.

这就是主动降噪的全部思路：造出一段与外来声音完全镜像的反噪声，并在同一时刻把它播进你的耳朵。

Listen, flip, play听、翻转、播放

A noise-cancelling earcup has a small microphone on the outside. It hears the noise a split second before the noise gets in. A chip flips that signal upside down and sends it to the speaker, which is already sitting right next to your ear.

降噪耳罩的外侧有一个小麦克风，它比噪声钻进耳罩早那么一瞬间就听到了噪声。芯片把这个信号上下翻转，送给紧贴在你耳边的扬声器。

This is why the effect feels so strange the first time: the headphones are not blocking anything extra. They are adding sound, and the sum happens to be quieter.

这也是第一次戴降噪耳机时感觉如此奇怪的原因：耳机并没有额外挡住什么，它是在添加声音，而加起来恰好更安静。

Timing is everything时机决定一切

There is a catch. Listening, flipping and playing all take time. The anti-noise always arrives a tiny bit late, and "a tiny bit late" matters a lot more for some sounds than for others.

但有个问题：听、翻转、播放都要花时间。反噪声总会晚到一点点，而“晚一点点”对有些声音影响很大，对另一些几乎没影响。

A fixed delay is a small fraction of a slow wave's cycle but a big fraction of a fast one's. With 60 microseconds of delay, a 100 Hz hum is cancelled almost perfectly, while anything above about 2.8 kHz would be made worse. The crossover sits where the delay equals one sixth of a cycle. That is why the electronics are built to be extraordinarily quick, and why real headphones fade their anti-noise out at high frequencies instead of risking it.

同样一段延迟，对慢波来说只占周期的一小部分，对快波来说却占了一大截。延迟 60 微秒时，100 Hz 的嗡嗡声几乎被完全抵消，而 2.8 kHz 以上的声音反而会被放大。分界点就在延迟等于六分之一个周期的地方。所以降噪电路要做得极快，真实的耳机也会在高频处逐渐减弱反噪声，而不是去冒这个险。

What you actually hear你实际听到的

Real noise is not one tone. It is a mixture of many, from the deep drone of an engine to the hiss of air and the chatter of voices. Two different mechanisms split the work:

真实的噪声不是单一音调，而是许多声音的混合：从引擎低沉的轰鸣，到气流的嘶嘶声，再到人声的嘈杂。两种不同的机制分工合作：

Cushions and a snug seal are good at stopping high frequencies, which is exactly where the electronics struggle. The electronics are good at deep, steady drones, which pass straight through foam. Together they cover the whole range. It is also why you can still hear someone talking to you on a plane: speech lives in the middle, where neither trick is at its best.

耳垫和贴合的密封擅长挡住高频，而这正是电路的短板；电路擅长对付低沉、稳定的轰鸣，而这种声音能直接穿透泡棉。两者合起来覆盖了整个频段。这也解释了为什么在飞机上你仍然听得见有人跟你说话：人声落在中间，两招都不是最强的地方。

Headphones that learn会学习的耳机

The outside microphone can't know exactly how the noise changes on its way into your ear: your head, the fit of the cups and even your hair all reshape it a little. So most headphones also have a second microphone inside the cup. It listens to whatever is left over and keeps nudging the anti-noise until the leftover is as small as possible.

外侧麦克风无法确切知道噪声进入耳朵的路上会怎么变：你的头型、耳罩的贴合度，甚至头发，都会让它稍有变化。所以大多数耳机在耳罩内侧还有第二个麦克风。它听的是剩下的声音，并不断微调反噪声，直到剩余部分尽可能小。

Each nudge follows a simple rule: if the leftover still looks like the noise, add a little more of the mirror image; if it looks like the anti-noise, back off. Small steps are slow but steady. Big steps adapt quickly, until they start overshooting. Engineers spend a lot of effort finding the sweet spot, because the noise around you keeps changing as you move.

每一次微调都遵循一条简单规则：如果剩余声音还像噪声，就多加一点镜像；如果它像反噪声，就退回一点。步子小，慢但稳；步子大，适应快，直到开始矫枉过正。工程师要花很大力气找到这个平衡点，因为你一移动，周围的噪声就在变。

Everything else is the same trick其他功能都是同一招

"Transparency" mode runs the same hardware the other way: instead of cancelling the outside microphone's signal, it plays it through, so you can hear the announcement. Cars use anti-noise to quiet engine drone in the cabin. All of it rests on the idea from the second figure: the air adds pressures, so the right sound, at the right moment, adds up to silence.

“通透模式”用的是同一套硬件，只是反过来用：不抵消外侧麦克风的信号，而是把它放进来，让你听得见广播。汽车也用反噪声来压低车厢里的发动机轰鸣。这一切都建立在第二张图的道理上：空气会把压强相加，所以在对的时刻加上对的声音，结果就是寂静。