Sound Wave Interference

Two loudspeakers in a Blender-built room: drag the microphone through the loud and silent bands of the pressure field; path difference, phase, level and far-field angles update live.

Switch to 2D

Presets

Playback

Sources

Microphone and listening

Camera

Layers

Quality

Model check

The science

What is modelled

Two loudspeakers on stands radiate the same tone into a room treated as free space at ear height (1 m). Each is a point source of spherical waves whose pressure falls as 1/r. At every spot the two waves add: where the crests arrive together the sound is loud (constructive interference), where a crest meets a trough it is quiet (destructive). The colours show time-averaged intensity, or with the layer switched on the instantaneous pressure (orange positive, blue negative).

Equations

  • Pressure from source i: pi = (Ai/ri) cos(ωit − kiri + φi), with k = ω/c = 2π/λ and c ≈ 331.3 + 0.606 T m/s.
  • Same frequency: the sum has complex amplitude P = a1e−ikr1 + a2e−ikr2+iφ0 and time-averaged intensity I = |P|²/2 = (a1² + a2² + 2a1a2 cos δ)/2, with δ = k(r1 − r2) + φ0.
  • Path difference Δ = r1 − r2. For equal phases maxima sit where Δ = mλ and nulls where Δ = (m + ½)λ. Far from the pair Δ = d sinθ, so the loud directions satisfy d sinθ = (m − φ0/2π) λ: about 2d/λ + 1 of them, only the centre line when d < λ.
  • Complete cancellation needs equal amplitudes; with B/A below 1 the nulls are shallow, with minimum intensity (a1 − a2)²/2.
  • Detuned sources (Δf > 0): δ(t) = δ0 + 2πΔf t, so the pattern slides and the level at a fixed point pulses at the beat frequency Δf.

Checks run in the page

  • A brute-force scan of the exact two-source sum at 2 km finds the intensity maxima and compares them with d sinθ = (m − φ0/2π)λ.
  • The mean of p(t)² over exactly one period at the microphone is compared with the phasor formula |P|²/2.
  • The wavelength measured from zero-crossings of speaker A alone is compared with c/f; the linear speed-of-sound formula is compared with √(γRT/M).

Simplifications

No room reflections (real rooms add echoes and standing waves; the foam panels suggest an anechoic-style space), point sources instead of cabinets with directivity, no air absorption, linear small-amplitude acoustics. The cone motion in the picture is exaggerated and the wave pictures run in slow motion (factor shown); the graphs and the optional sound use real time. Level is relative to speaker A alone at the same point, so no absolute calibration is needed. The listening button synthesises what a microphone at that spot would record: two oscillators, each delayed by r/c and weighted by 1/r, so the cancellation really happens in the sound.

What to try

  • Press Mic to a quiet spot, then choose Anti-phase: the null moves to the centre line.
  • Raise the frequency: the bands get narrower; lower it until d < λ and a single broad lobe remains.
  • Set speaker B to 0.4 loudness: the nulls fill in.
  • Detune B by 2 Hz and press Listen: the level pulses twice a second.
Loading model...
Loading Blender model...
Level along the mic line: coherent vs incoherent
Pressure at the microphone (A, B and the sum)
Far-field intensity vs angle
Made with 🟠 Blender · Three.js