What Is Phase in Audio, and What Is Phase Cancellation?

Sound travels as a wave: air pressure rises above normal, falls below it, and rises again, over and over. One full rise-and-fall is a cycle, and phase describes where in that cycle the wave is at a given moment, usually in degrees from 0 to 360. On its own, a single sound's phase is inaudible. Phase matters when two copies of a similar sound meet. If their peaks line up, they reinforce each other and get louder. If the peak of one meets the trough of the other, they push in opposite directions and partly or completely cancel.

Imagine two identical waves, one shifted by half a cycle (180 degrees). Every time one is at its highest, the other is at its lowest, so adding them gives zero. That is full phase cancellation, and Wikipedia's article on out-of-phase stereo describes exactly this: two identical but inverted waveforms summed together cancel each other out. In real recordings the copies are rarely identical. One microphone is a little further away, a reflection arrives a few milliseconds late, or a channel has been processed differently. The result is partial cancellation: some frequencies drop, others stay or even get louder. The sound becomes thin, hollow or "phasey" rather than silent.

When a sound is mixed with a slightly delayed copy of itself, the delay cancels certain frequencies and reinforces others. On a frequency graph the result looks like the teeth of a comb, which is why it is called comb filtering. The notches are evenly spaced, and the delay sets the spacing. As Wikipedia's comb filter article works out, a 10 ms delay gives notches every 100 Hz, starting at 50 Hz, then 150 Hz, 250 Hz and so on. A shorter delay spreads the notches further apart. You hear comb filtering when a microphone picks up a voice directly and again off a nearby desk, when two mics capture the same instrument at different distances, and when two speakers play the same sound to a listener sitting closer to one of them.

The two words are often mixed up, and many "phase" buttons are really polarity switches. Polarity inversion flips a waveform upside down: every positive value becomes negative and vice versa. It affects all frequencies equally and happens instantly. A polarity flip is what makes two identical signals cancel completely. Phase shift moves a waveform in time, or changes how different frequencies are timed relative to each other. Because a fixed time delay is a different fraction of a cycle at each frequency, a delay causes cancellation at some frequencies and reinforcement at others, which is comb filtering. So "phase inversion" usually means a polarity flip. It fixes problems where one signal is upside down, for example a microphone under a snare drum facing the opposite direction from the one on top, but it does not fix a timing difference.

Multiple microphones on one source. Two mics at different distances from a guitar amp, drum or piano receive the sound at slightly different times. Mixed together, they comb filter. Top and bottom mics. A mic under a snare drum sees the drum head moving the opposite way from a mic on top, so the two are opposite in polarity. Reflections. A microphone near a hard surface picks up the direct sound plus a delayed reflection. Wide stereo effects. Some stereo widening, chorus and reverb effects create left and right channels that partly cancel when summed to mono. Wiring and processing errors. A miswired cable or a plugin set to invert one channel can flip polarity without anyone noticing.

Listen in mono. Switch your monitoring to mono or sum the channels: if instruments get noticeably thinner or quieter, or a vocal drops back, part of the mix is out of phase. Many phones, smart speakers and venue systems play in mono, so this matters for real listeners. Flip the polarity and compare. On a multi-mic recording, invert one mic and listen. The setting that sounds fuller and has more low end is usually the correct one. Look at the waveforms. Zoom in on a sharp sound such as a drum hit on two tracks. If one is shifted later or upside down relative to the other, you have found the cause. Use a correlation meter. Many DAWs and metering plugins show stereo correlation: values near +1 are mono-safe, while values near or below 0 warn of cancellation.

1. Fix it at the source. Move microphones until they sound full together. A common rule of thumb, the 3:1 rule, places each microphone at least three times as far from the other mics as it is from its own source, which keeps the spill quieter than the direct sound. 2. Flip polarity where one signal is upside down, such as a snare-bottom or kick-inside mic. 3. Time-align the tracks. Nudge the later track earlier until the waveforms line up, or use an alignment plugin that does it automatically. 4. Use fewer microphones. If two mics fight each other, one well-placed mic often sounds better than two. 5. Reduce reflections. Move away from hard surfaces or add absorption near the microphone. 6. Tame stereo effects. Narrow over-wide reverb or chorus, or check them in mono before committing.

Yes. Many subwoofers have a phase switch (0 or 180 degrees) or a variable phase control. Because the subwoofer usually sits in a different place from the main speakers, its sound can arrive out of step with theirs around the crossover frequency, which thins the bass where they overlap. Play music or a bass test tone from your listening seat, try each setting, and keep the one where the bass sounds fullest and most even. Moving the subwoofer can matter just as much as the switch.

Phase cancellation can also be a tool. Out-of-phase stereo, or OOPS, inverts one stereo channel and sums the two. Anything identical in both channels, usually the centred lead vocal, cancels, while sounds panned to the sides survive. Wikipedia notes it has long been used for crude karaoke tracks and even to reveal material hidden in some stereo mixes. The results are rough, because bass, kick and snare are often centred too and drop out with the vocal. Noise-cancelling headphones use the same principle, generating an inverted copy of outside noise so the two cancel at your ear.

Mono compatibility means a stereo mix still sounds right when its channels are summed into one. A mix with phase problems can lose its vocal, its bass or its width when played on a single smart speaker or a phone held at arm's length. Before releasing music, a podcast or a video, listen once in mono. To hear exactly what mono playback will do, run a copy of your file through our stereo to mono tool and compare it with the original. If a mono recording plays in only one ear, mono to stereo copies it to both channels without changing the sound. Our guide to stereo vs mono covers when each makes sense.