11. Loudspeaker & Electroacoustics · Concept 9 of 11
Group Delay
A measure of how some frequencies come out of a speaker slightly later than others, smearing the timing.
A relay race: treble crosses on time, bass lags by tau_g milliseconds, smearing the punch.
What it is
Group delay is how much later some frequencies leave a speaker than others, smearing timing and softening transients.
Key facts
Definition: group delay = how long a band of frequencies is delayed; the negative slope of the phase-vs-frequency curve.
Formula: tau_g(f) = -(1 / 360) x (d phi / d f), where tau_g = group delay in seconds, phi = phase in degrees, f = frequency in Hz.
Units: seconds, but in pro audio always milliseconds (ms). 1 ms = 1/1000 s.
Speed of sound in air = 343 m/s at 20 C; 1 ms of delay = 0.343 m = 34.3 cm of path difference.
Audibility (Blauert and Laws): the ear is most sensitive around 2 kHz (~1 ms threshold), with ~1.5 ms at 4 kHz, ~2 ms at both 1 kHz and 8 kHz, and ~3.2 ms at 500 Hz, rising further in deep bass; sensitivity is a U-shaped curve, so bass smear is forgiven far more than midrange smear.
Steeper filter = more group delay: a 24 dB/oct (4th-order Linkwitz-Riley) crossover smears more than 12 dB/oct (2nd-order); a 30-40 Hz sub high-pass can add 10-20 ms at the bottom octave.
FIR (linear-phase) filters give FLAT group delay = zero smear but cost fixed latency (often 5-20 ms); IIR (analog-style) filters are cheaper, lower latency, but always add some group delay.
Doubling power or sources = +3 dB; doubling SPL pressure = +6 dB; -3 dB = the half-power point used to spec crossover/bandwidth; +10 dB needed to sound twice as loud.
Excess group delay = smear left AFTER subtracting pure propagation time; that is the part you hear as 'loose' or 'woolly'.
Ported (bass-reflex) subs ring at their tuning frequency and have MORE group delay near tuning than sealed subs.
How it works
A speaker's phase shifts with frequency, not by the same amount everywhere.
Where phase changes fast with frequency, those bands arrive late = high group delay.
Crossovers and high-pass filters bend the phase curve hardest near their corner frequency.
Low frequencies physically take longer to develop, so subs sit a few ms behind the tops.
Stack enough delay and a kick's attack arrives smeared across several ms instead of one sharp hit.
Fix it with sub alignment delay, gentler filter slopes, or a linear-phase (FIR) processor.
Real examples
Kick drum through a 4th-order 80 Hz crossover sounds 'woolly' because the sub band lags the punch by several ms.
Ported sub ringing at its 35 Hz tuning leaves bass notes hanging instead of stopping clean.
Tops time-aligned but subs not delayed: the slap of a snare arrives before its body = smeared transient.
Switching a system processor from IIR to FIR linear-phase tightens the low end with no EQ change.
A steep 30 Hz infrasonic high-pass adds 15 ms group delay and makes deep bass feel sluggish.
How it helps in live sound
Measure it: use Smaart or REW, view the group-delay (or phase) trace, watch the crossover region.
Add sub-to-top alignment delay (typically 2-15 ms) so the sub arrives WITH the tops, not before.
Confirm alignment with a phase trace overlay: the two boxes should sit on top of each other through crossover.
Prefer 24 dB/oct LR crossovers but know they cost group delay; drop to 12-18 dB/oct if the low end feels loose.
Use FIR/linear-phase presets on DSP (Lake, Galileo, d&b ArrayProcessing) for tight kick at the cost of a few ms latency.
Sealed subs over ported when transient tightness matters more than maximum low-end extension.
Everyday analogy
It is a relay race where the bass runner starts a beat after the treble runner, so the team crosses the line out of step and the punch is lost.
Watch out
Myth: 'flat frequency response means perfect timing.' Wrong: a system can be flat in level yet have heavy group delay, so it measures flat but still sounds smeared.
Fun fact
Linear-phase FIR filters can give literally zero group delay variation across the band, but they 'pay' for it with pre-ringing and a fixed whole-system latency you cannot remove.
Key takeaways
Group delay = some frequencies leave the speaker late, smearing timing.
It is the slope of the phase curve: tau_g = -(1/360)(d phi / d f).
1 ms of delay = 34.3 cm of sound travel at 343 m/s.
Ears tolerate ms of bass delay but spot a few ms up top instantly.
Steep crossovers and high-pass filters are the main cause; FIR fixes it for latency cost.
Align subs to tops and watch the phase trace to keep kicks punchy, not mushy.