Comb Filtering and Reflection Delay: Why Tiny Path Changes Matter

Comb filtering occurs when a signal combines with a delayed related version, producing repeated frequency reinforcement and cancellation. For an ideal equal-level, same-polarity delayed copy with delay τ seconds, notches occur at (2k + 1)/(2τ), and adjacent notches are spaced by 1/τ.

A reflection can supply that delayed copy. The exact depth of real-room notches depends on reflection level, phase, spectrum and additional paths. The simple formula is a controlled model, not a complete prediction of a furnished room or a microphone recording.

Convert path difference to delay

A reflected route is usually longer than the direct route. Divide the extra distance by sound speed to obtain the arrival delay relative to the direct sound. With an assumed 343 metres per second, an extra path of 0.686 metres gives 0.002 seconds, or two milliseconds.

Use the extra path, not the complete distance from speaker to listener. If the direct route is two metres and the reflected route is 2.686 metres, the relevant difference remains 0.686 metres. Confusing total distance with extra distance gives the wrong interference spacing.

The early reflections guide explains arrival geometry. The room acoustics calculator supports entered geometry calculations, but it does not automatically trace every speaker-desk-microphone reflection path.

An original two-millisecond example

For τ = 0.002 seconds, the first notch is 1/(2 × 0.002) = 250 Hz. The next is 3/(2 × 0.002) = 750 Hz, followed by 1250 Hz. The spacing between adjacent notches is 1/0.002 = 500 Hz.

Ideal same-polarity delayed-copy cancellation frequencies
DelayFirst notchNext notchNotch spacing
1 ms500 Hz1500 Hz1000 Hz
2 ms250 Hz750 Hz500 Hz
4 ms125 Hz375 Hz250 Hz

Increasing the delay brings the notches closer together in frequency. It does not simply make one fixed notch deeper. A different reflection geometry can therefore change the ripple pattern even if the reflected level stays similar.

The reverb calculator offers musical timing starting points, not an automatic comb-filter analyzer. A tempo-derived pre-delay belongs to a creative effect workflow. It should not be mistaken for a measured physical reflection delay.

Notch spacing and notch depth are different

Let the direct amplitude be one and the delayed-copy amplitude be a. At a cancellation frequency in the simplified same-polarity model, the residual magnitude is |1 − a|. At reinforcement, it is 1 + a. Complete cancellation requires equal magnitudes and the relevant opposing phase.

If a = 0.5, the ideal minimum is 0.5 and maximum is 1.5 relative to the original amplitude. Their ratio is three, or approximately 9.54 dB. The minimum relative to the direct signal alone is approximately −6.02 dB; these are different reference comparisons.

Real surfaces are frequency-dependent, and several reflections overlap. Do not infer a single reflection’s level from one notch without checking the model. A polarity inversion would also interchange the ideal reinforcement and cancellation conditions, so the stated polarity assumption matters.

Why reverb time is not enough

RT60 summarizes a decay rate under appropriate conditions. It does not uniquely identify an early path delay. Two rooms or positions can have similar decay estimates while differing in desk reflections, speaker-boundary paths and comb-filter patterns.

Read the RT60 definition and room acoustics overview before combining those descriptions. A narrow cancellation problem should not automatically be labeled an excessively long reverberation time.

Likewise, adding absorption area in a prediction does not specify which reflected path is reduced. The absorption coefficient guide distinguishes a material input from the actual position and direction of a reflection.

Investigate the path with controlled changes

  1. Check for duplicate digital paths before assuming a room reflection.
  2. Document speaker, microphone and nearby surface positions.
  3. Change one relevant distance and compare the frequency pattern.
  4. Inspect an impulse response for a plausible delayed arrival.
  5. Compare both channels and retain the same measurement settings.

A delayed duplicate track can create similar mathematics without any room at all. Check routing, plugin latency compensation and parallel processing. The relevant cause might be a dry copy mixed with a delayed path rather than an acoustic boundary.

The send-versus-insert guide helps distinguish intentional wet routing from duplicate dry paths. The pre-delay-versus-decay article separates onset offset from tail length when testing an effect.

Do not assume boosting a deep cancellation will solve it throughout a room. The direct and reflected components can both increase, and another listening position can behave differently. The purpose of the simple model is to identify a testable mechanism, not prescribe a universal equalizer setting.

For a check on the distance arithmetic, four milliseconds corresponds to an extra path of 1.372 metres at the stated sound speed. Halving that extra path halves the delay and doubles the ideal notch spacing. This relationship concerns the path difference, not necessarily a single speaker-to-wall distance.

Frequently asked questions

Is comb filtering the same as an echo?

They can arise from related delayed paths, but the perceptual result depends on delay, level and material. Comb filtering describes the frequency response created by interference, not only a separately heard repeat.

Why do the notches move when I move?

Your movement can change the difference between direct and reflected path lengths. That changes the relative delay and therefore the predicted interference pattern. Several reflections can change at once.

Does every reflection cancel completely?

No. Complete cancellation in the simple model requires matching amplitudes and the appropriate phase. Real reflections usually have different levels and frequency-dependent behavior, so the pattern is more complicated.

Can a reverb plugin create comb filtering?

Yes, certain early reflections, delays or parallel paths can create interference. That does not imply every algorithm has the same response or that its decay-time control determines notch spacing.

Is the first notch always 250 Hz?

No. That value belongs to the two-millisecond illustration with stated assumptions. Change the delay, polarity or signal relationship and the result changes.

Can this article measure my reflection delay?

No. The examples are calculated models. An actual path investigation needs appropriate timing or acoustic measurements and a check that the observed pattern fits the proposed cause.

Scroll to Top