🎛️ Comb Filtering — Audio Engineer’s Perspective
Comb filtering is a spectral interference phenomenon that arises when a signal is summed with a time-displaced replica of itself (typically 0.1–10 ms delay). The superposition produces a series of periodic cancellations and reinforcements across the frequency spectrum, yielding a response that visually and sonically resembles a comb.
What’s actually happening:
Phase-dependent interference
Frequencies whose wavelengths align out of phase undergo destructive interference (nulls), while others align in phase and are constructively amplified (peaks).
Not EQ — geometry in time
Unlike equalization, comb filtering is delay-governed, not gain-governed. You cannot “EQ it away” without reintroducing instability elsewhere.
Spectral spacing logic
The frequency distance between notches equals 1 / delay time.
Example: a 1 ms delay ⇒ notches every 1 kHz.
Why engineers care (and fear it):
Hollow / metallic timbre
The ear interprets dense notches as phasiness, nasality, or a “flanged but static” coloration.
Mono summing disasters
Stereo widening tricks (Haas delays, dual mics) often implode into comb filtering when collapsed to mono.
Live sound & acoustics
Reflections from walls, floors, or misaligned PA arrays generate time offsets that erode clarity—especially speech intelligibility.
Multi-mic setups
Poorly spaced microphones on the same source (drums, guitar cabs) invite inter-mic delay conflicts.
Common engineering triggers:
Parallel processing without latency compensation
Stereo tracks with micro-delay offsets
Close reflections in untreated rooms
Duplicate tracks nudged slightly in time
Delay thresholds (intuition guide):
<1 ms → severe coloration
1–5 ms → pronounced combing
5–20 ms → flanging / chorusing territory
>20 ms → perceptual separation begins
In one sentence:
Comb filtering is the spectral scar left by time misalignment—an acoustic reminder that sound obeys geometry as ruthlessly as physics.
🔧 Align first. Sweeten later.
January 29, 2026 435 1
