Speaker phase and time alignment: when timing matters as much as tone

In the octave around your crossover, two different speakers reproduce the same notes at once — your mains and your subwoofer. Whether those two copies reinforce each other or cancel depends on something no tone control can see: when each one arrives at your ears. That is speaker phase and time alignment, and in the handoff region it matters exactly as much as tone, because there, timing is tone.

It is also one of audio’s most confused subjects, because three different things — delay, phase, and polarity — all get called “phase,” and one of them is a switch on the back of your sub that promises more than it can deliver. So first the vocabulary, then where timing errors come from, and which of them you can actually hear.

Time alignment, phase, and polarity are not the same thing

Delay is arrival time. Sound covers about 1,125 feet per second — call it a foot per millisecond. A speaker three feet farther from you than its partner is about 2.7 ms late, and that lateness is identical at every frequency. Delay is one number, measured in time.

Phase is where a wave is in its cycle at one frequency, in degrees. The same delay produces a different phase shift at every frequency: an 80 Hz cycle takes 12.5 ms, so a 2 ms timing error shifts it roughly 60 degrees, while at 1 kHz — where a cycle lasts just 1 ms — that same 2 ms is two full revolutions. One timing error, a different phase shift at every frequency. The two views describe the same physics from different angles.

Polarity is the control labeled “phase” on many subwoofers, and it is neither of the above. Flipping it turns the waveform upside down — a 180-degree change at every frequency at once. That is not alignment; it is a coin with two faces. If your sub sits 120 degrees out at the crossover, the flip leaves it 60 out: better. If it was 60 out, the flip makes it 120: worse. Neither position is aligned. You are choosing the smaller error.

One more term, because your measurement software will show it: group delay. It plots how much each frequency band is delayed relative to the others. A perfect system delays everything equally; real crossovers and resonances delay some bands more, so the low half of a kick drum can lag its own beater click. Small amounts are inaudible. Large amounts at low frequencies are part of what people mean when they call bass “slow.” It is not a separate phenomenon — just delay, plotted per frequency — and it is one of the more honest views of what a system does to timing.

Where phase and time alignment problems come from

Inside each speaker. On a flat baffle, the tweeter’s acoustic center sits an inch or so closer to you than the woofer’s, because the woofer’s voice coil is deeper in the cabinet. An inch is 74 millionths of a second, which sounds like nothing — until you notice that at a 2.5 kHz crossover a full cycle lasts only 0.4 ms, so that inch is about 66 degrees. The crossover filters add frequency-dependent delay on top; that is simply what filters do. Designers manage both inside the cabinet, and the speaker you bought includes their compromises.

Between speakers. Every foot of distance difference is about a millisecond. An off-center desk, one monitor pushed nearer a wall, a surround speaker mounted closer than its twin — each shows up as an arrival offset before any signal processing gets a say.

The subwoofer. Subs get placed where bass works best, which is rarely the same distance from you as the mains. A path difference of four to seven feet is ordinary — 3.5 to 6 ms, or 100 to 180 degrees at an 80 Hz crossover. The sub’s own low-pass filter delays its output further still. Geometry and electronics stack.

The room itself. Room modes are resonances: they store a note’s energy and release it after the note stops, so the room keeps playing after every driver has gone still. That is a timing problem of a different kind — not a wrong arrival but a departure that never quite finishes — and no arrival-time adjustment touches it. The mechanism, and what actually helps, is covered in our piece on room modes.

What misalignment sounds like — and where it bites hardest

Honesty matters here, because phase has a reputation problem in both directions: folklore treats it as everything, and skeptics dismiss it as nothing. The truth splits cleanly in two.

The subtle case is the phase response of a single speaker playing alone. Its audibility is real but modest and heavily program-dependent: sparse, transient material — a click, a dry drum hit — can expose it; dense mixes mostly mask it. If a product promises a night-and-day transformation from phase correction of an already decent speaker, be skeptical.

The not-subtle case is two sources overlapping. Around the crossover, sub and mains produce near-equal output by design — that is what a crossover is. Two equal signals 180 degrees apart cancel. An 80 Hz wave is about 14 feet long (4.3 m), so a seven-foot difference in path length puts sub and main half a wave apart — and a normal room offers seven feet of asymmetry without trying. The result is a null centered on the crossover: a measurable hole in the frequency response, in exactly the region you bought a subwoofer to strengthen. Nothing psychoacoustic about it. It shows up on any measurement, and it sounds like the bottom fell out of the low end.

Partial misalignment costs too, just more quietly. Two equal sources in step sum to +6 dB. At 90 degrees apart they manage +3 dB. At 120 degrees, the pair plays no louder than one source alone — the sub is drawing power and moving air to add exactly nothing. And between crossover misalignment and modal ringing, bass transients smear: the impact arrives in pieces instead of at once.

EQ cannot fill the null, for the same reason it cannot fill a modal null: boosting the crossover region feeds both sources, and they keep subtracting. The problem is timing, so the fix is timing.

Why you can’t align a subwoofer by ear — or with one knob

A crossover null does not sound like a timing problem. It sounds like missing bass — which reads as a level problem, so you turn the sub up. But level cannot restore what cancellation removes: raise the sub 6 dB into a 180-degree null and the sum crawls back only to about the level of the mains alone, while everything around the null gets louder. The ear diagnoses the wrong ailment, and the treatment brings new symptoms.

The deeper problem: the timing relationship is not a property of your equipment. It is a property of each listening position. Move one seat along the couch and both path lengths change by different amounts — a seat three feet closer to the sub than its neighbor shifts the sub-to-main relationship by about 2.7 ms, close to 77 degrees at 80 Hz. The knob on the sub applies one setting to the whole room. It can genuinely align one point; the geometry guarantees it cannot align them all. Whatever you dial in by ear from the chair, you dialed in for the chair.

Low frequencies make it harder still: what reaches your ear at 80 Hz is the direct sound overlapped with the room’s reflections, so there is no single clean “arrival” to line up by feel. A measurement resolves what the ear cannot: a sweep played through each speaker yields its impulse response at each microphone position — a timestamped record of when and how that speaker’s sound arrives there. (Our monitor calibration walkthrough covers how to read one.) Add more subs and the number of relationships multiplies — every sub-to-main and sub-to-sub pair, at every seat — which is why multi-sub setups are exactly where guess-and-check breaks down.

What a solver can do with measured timing

Once every arrival is measured, nothing about timing has to be assumed — and that changes what correction can be. Instead of one delay per speaker tuned for one seat, alignment becomes an optimization problem: find the set of per-speaker delays and filters that makes the whole system sum correctly across all the measured positions at once.

That is how Perfect Soup, our calibration software, treats it. It measures each speaker and subwoofer from 30 to 100 positions around the listening area, then solves everything jointly — per-speaker timing, the phase handoff at each crossover between a main speaker and its subs, and the frequency response — as one GPU-accelerated optimization across every speaker and every position simultaneously. A sub shared by several channels gets one coherent correction rather than a different answer per channel. Nothing in the solve is assumed: not the distances, not the crossover behavior, not what the room does to each arrival. It is all read from your measurements, and the result loads as a plugin on your monitor bus.

The limits: a correction aligned across an area is a solved compromise — the best available agreement between seats, not perfection at each one. And no filter moves a speaker. If the layout is grossly asymmetric, a tape measure is still the first tool; the solver makes a good setup right, it does not make a wrong setup good.

If your low end has never quite locked in — sub clearly working, bass somehow thin — timing is the first suspect, and it stays invisible until you measure it. Here’s how Perfect Soup measures a room and solves the alignment.