Time Alignment Using Group Delay

These are the tools required. Everything here is either free or a one-time purchase you will use for every tune.

• 01
  REW (Room EQ Wizard) — Free download at roomeqwizard.com. This is your measurement platform for the entire process.
• 02
  Calibrated measurement microphone — Dayton EMM-6, miniDSP UMIK-1, or equivalent. Must include a .cal calibration file. An uncalibrated mic gives you wrong numbers.
• 03
  USB audio interface — Focusrite Scarlett 2i2, MOTU M2, or similar. Provides clean I/O and loopback capability for the baseline measurement.
• 04
  Microphone stand or mount — You cannot move the mic between driver measurements. Mark the position with tape if needed.
• 05
  Your DSP unit — Any processor with per-channel delay entry in milliseconds. Arc Audio, miniDSP, Helix, Mosconi, and most others support this.

This is the step most people skip, and it creates problems later. Time alignment delay should be the last major step in your tune, not the first.

Before you take the measurements this guide uses, each driver should already be tuned to its individual target response. That means:

1. Crossover points set correctly for each driver
2. Individual EQ applied to each driver to hit its target curve
3. Gain structure set and stable

Once each driver is sounding correct individually, you are ready to align them together. That is what the rest of this guide covers.

Before measuring any speaker, run a loopback measurement. This confirms your signal chain is clean and gives you a flat baseline with an impulse that peaks right at t=0. You will use this to verify your alignment is correct at the end.

How to run a loopback in REW:

1. Connect a short cable from your interface output directly to its input
2. In REW, open the Measurement window and select SPL and Phase
3. Run the sweep. You should see a flat line close to 0 dB from 20 Hz to 20 kHz
4. The impulse response should peak at exactly t=0 with no pre-ringing

With the mic locked in position at the listening seat, measure each driver one at a time. Mute all other drivers through the DSP before each measurement.

1. Tweeter — mute midrange and subwoofer
2. Midrange — mute tweeter and subwoofer
3. Subwoofer — mute tweeter and midrange

The most common time alignment mistake: aligning the peaks of the impulse responses. It seems logical. It is wrong.

The peak of an impulse response is not determined by when the sound arrives. It is determined by the highest frequency present in that driver's passband at the highest level. A tweeter passes 20 kHz, so its peak lands near t=0. A subwoofer only passes energy up to 80 Hz, so its peak lags by several milliseconds even when the physical distance is the same. That lag is a frequency-related artifact, not a distance measurement.

You can sometimes use the point where the impulse begins (not the peak) for the tweeter and midrange. But for the subwoofer, that starting point is nearly impossible to read reliably. It's a guessing game. The group delay plot removes all of that uncertainty.

With all three driver measurements overlaid in REW, click the GD tab (group delay). Set the Y-axis to 0 to 15 ms and the X-axis to 20 Hz to 20 kHz.

Each driver's curve will be flat within its passband and noisy outside of it. Ignore the noisy regions. The tweeter's curve will be noisy below its crossover point. The subwoofer's curve will be noisy above 200 to 300 Hz. That is expected.

You can also use the phase plot to read alignment. Match the phase slope of each driver at the crossover point. Group delay tends to be easier to read because the values are already in milliseconds, but both methods work.

In this example, the subwoofer is farthest away and has the most delay. It is the reference driver. The tweeter and midrange need delay added to match it. The sub itself gets 0 ms of added delay.

Use the sliders below to enter the group delay values you read from REW. The calculator shows exactly what to enter into your DSP for each channel.

After entering your values, verify the work: In REW, take a new measurement with all three drivers active and sum them. Compare that summed measurement to your original loopback. The frequency response should be flat with no dips at the crossover frequencies. If you see dips at 80 Hz or 3 kHz, the timing is still off and needs adjustment.