Phase is timing with consequences

You are mixing a snare drum with two microphones, one on the top skin and one on the bottom. The snare sounds thin, lacking the explosive weight it needs to drive the song. Your first instinct is to load up an EQ. You boost the low-mids at 200 Hz. You push the gain up by 6 dB. But instead of sounding thick, the snare gets even more hollow.

You are throwing EQ boosts at a physical void. If the two microphones are out of phase, their waveforms are fighting each other. Stacking EQ boosts on a canceled frequency does not work. You are boosting signals that subtract from each other upon combination, which just wastes headroom.

Phase is a timing relationship. When a sound wave hits two microphones at slightly different times, the waveforms do not align. When you sum these channels to mono, the peaks of one wave can meet the troughs of the other. This causes destructive interference, which pulls down the volume of specific frequencies.

This cancellation is most noticeable in the low end, where waveforms are wider and easier to disrupt. If your kick drum and sub-bass are out of phase, they will fight for space. Every time the kick hits, the sub-bass will pull its low frequencies down, leaving you with a weak foundation that falls apart on club sound systems.

A waveform is cyclical, and phase represents the angle of a wave at a specific point in its cycle. We measure this angle in degrees, from 0 to 360, or in radians.

Destructive interference is strongest when two waveforms are 180 degrees ($pi$ radians) out of phase. In this state, they are moving in opposite directions. The mathematical sum of two identical waves offset by half a wavelength is zero:

$$y(t) = x(t) + x\left(t - \frac{T}{2}\right) = 0$$

In this equation, $T$ represents the period of the wave. If the waves are not identical, or if they are offset by a different phase angle, the cancellation is partial. This partial cancellation is what causes comb filtering, which creates a series of steep notches across the frequency spectrum.

You can hear phase cancellation clearly by creating a manual cancellation test in your session.

A common mistake is believing that all tracks must be perfectly in phase. This is physically impossible on multi-mic setups. If you align a drum room mic to be perfectly in phase with the snare, it will be out of phase with the tom-toms.

Furthermore, minor phase differences are what create stereo width. If you run a delay plugin to create a wide stereo image (using the Haas effect), you are using phase delay to trick the brain.

Another error is ignoring the mono check. A mix can sound wide and lush in stereo, but when played on a mono phone speaker or a club system, the phase-canceled tracks will disappear entirely.

Check your phase relationship before you touch an EQ. When mixing multi-mic setups, sum the channels to mono and toggle the polarity of one track. If the low end returns, keep the polarity flipped. Use a utility plugin to nudge the timing of the delayed track in millisecond increments until the low end locks together. Always check your master output in mono to verify that your lead vocals and low-end instruments do not disappear.