Haas Effect

A second copy under ~40 ms fuses into one sound, located at whichever arrives first.

What it is

If a second copy of a sound arrives within about the first 40 ms, you hear one fused sound, not an echo.

Key facts

• Fusion window: roughly 1-40 ms delay = heard as ONE sound, not two.
• Localisation locks to the FIRST-arriving wavefront, even if the later copy is up to ~10 dB LOUDER.
• Echo threshold: past ~40-50 ms (speech) you start hearing a distinct echo / slap.
• Music tolerates longer (~50-80 ms) before splitting; clicks/transients split as early as ~5 ms.
• Speed of sound in air = 343 m/s at 20 C (about 1130 ft/s).
• Distance per ms of delay = 0.343 m (34.3 cm) per millisecond.
• Delay time formula: time (ms) = distance (m) divided by 0.343 (or distance in metres x 2.91).
• Below ~1 ms the two copies sum into comb filtering and image shift, NOT Haas fusion.
• +6 dB = doubling sound pressure (2x voltage); +10 dB = roughly 2x perceived loudness.
• -3 dB = half the power (half-power point); -6 dB = half the pressure.

How it works

1. Take a sound and a copy of it from a second source.
2. Delay the copy by about 1 to 40 ms.
3. Brain fuses the two into a single sound (precedence effect).
4. Perceived location snaps to whichever arrives FIRST.
5. Image can be pulled toward the first speaker even if the second is louder.
6. Push delay past ~40-50 ms and the copy splits off as an audible echo.

Real examples

• Mono vocal copied to a second speaker 15 ms late: image still snaps to the first speaker, not the centre.
• Stereo widener plugin: pan one side, delay the copy 10-25 ms to fatten a guitar or synth.
• Delay-tower / fill speaker timed so the audience still hears the sound as coming from the stage.
• Slap clap in a small tiled room: reflections under 40 ms make one big clap, not an echo.
• Doubling a DI bass with a 20 ms delayed copy to add width without obvious repeat.

How it helps in live sound

• Delay/fill towers: add (distance in metres x 2.91) ms so the fill fires AFTER the main, keeping the image on stage.
• Add a small extra 5-15 ms 'Haas offset' on top of physical delay so the audience localises to the stage, not the nearest box.
• Stereo width: duplicate a mono track, pan hard L/R, delay one side 10-30 ms (watch mono fold-down for comb filtering).
• Centre-cluster vocal louder but timed first keeps the vocal image centred over wide L/R PA.
• Keep all Haas delays under ~40 ms or you get slapback; check the worst seat, not just front-of-house.
• Speed of sound shifts ~0.6 m/s per degree C, so re-time delays if temperature swings a lot outdoors.