Sound Fundamentals · Wave Interference

How acoustic beats work: interference, envelope rate and perception

When two close frequencies combine, their shifting phase alignment creates a rhythmic rise and fall in amplitude. Understanding the difference between physical interference in air and central binaural processing clarifies what each rhythmic stimulus actually delivers to the ear.

How the sound works How it is perceived Page updated: 11 November 2026 · Literature search cutoff: 10 November 2026

1. How the sound works: Wave interference and beat rate

When two sinusoidal sound waves of slightly different frequencies f1 and f2 travel through the same medium or are summed in the same audio channel, their instantaneous pressures add together by linear superposition (Russell, 2011; Moore, 2012). Because the two waves oscillate at slightly different rates, their relative phase drifts continuously over time.

When the crests of both waves align (in phase), constructive interference produces a maximum combined amplitude. Half a beat cycle later, when a crest of one wave aligns with a trough of the other (180° or π radians out of phase), destructive interference reduces the combined amplitude toward a minimum. The rate at which the physical envelope rises and falls is determined by the absolute frequency difference:

fbeat = |f2 − f1|

For example, combining a tone at f1 = 211 Hz with an equal-amplitude tone at f2 = 221 Hz produces an envelope beat frequency of fbeat = |221 - 211| = 10 Hz (ten physical envelope maxima per second, spaced every 100 ms; under suitable listening conditions when both tones fall within a single cochlear critical band, this physical modulation is perceived as a 10 Hz loudness fluctuation).

Why a 5 Hz cosine factor produces a 10 Hz beat rate

By the trigonometric sum-to-product identity, the sum of two unit-amplitude cosines at 211 Hz and 221 Hz is:

cos(2π·211t) + cos(2π·221t) = 2·cos(2π·216t)·cos(2π·5t)

Or, when normalized by 0.5 to keep peak amplitude at 1.0:

x(t) = 0.5·[cos(2π·211t) + cos(2π·221t)] = cos(2π·216t)·cos(2π·5t)

Here, favg = (f1 + f2)/2 = 216 Hz is the frequency of the rapid cosine factor in the product representation (which completes 216 cycles and 432 zero-crossings per second), and fmod = |f2 − f1|/2 = 5 Hz is the frequency of the signed slow cosine factor cos(2π·5t). Because an ear responds to the unsigned envelope magnitude |cos(2π·5t)| regardless of whether the carrier phase is +1 or -1, one full 200 ms cycle of the signed 5 Hz factor contains two envelope maxima (at t = 0 ms where cos = +1, and at t = 100 ms where cos = -1) and two zero nodes (at t = 50 ms and t = 150 ms). Thus, the physical envelope repeats at 2 × 5 Hz = 10 Hz.

Crucially, in a long steady-state Fourier spectrum of x(t), the only spectral lines present are the two original frequencies f1 = 211 Hz and f2 = 221 Hz. In the trigonometric product form cos(2π·216t)·cos(2π·5t), the 216 Hz rapid cosine factor, the 5 Hz signed slow cosine factor, and the 10 Hz repetition rate of the unsigned envelope |cos(2π·5t)| are time-domain factorization and envelope properties, not separate third or fourth acoustic sine waves added to the air.

Mathematical details: Equal vs. unequal amplitudes and consistent normalization

When two tones have arbitrary amplitudes A1 and A2, their unnormalized linear sum is:

x(t) = A1·cos(2π·f1·t) + A2·cos(2π·f2·t)

Using phasor addition at the instantaneous phase difference Δφ(t) = 2π(f2 − f1)t, the analytic envelope magnitude E(t) is:

E(t) = √(A12 + A22 + 2·A1·A2·cos(2π·(f2 − f1)·t))

As cos(2π(f2 − f1)t) oscillates between +1 and -1, the unnormalized envelope fluctuates between a constructive maximum and a destructive minimum:

Emax = A1 + A2,     Emin = |A1 − A2|

If a two-tone mixer applies a 0.5 normalization factor, xnorm(t) = 0.5·[A1·cos(2π·f1·t) + A2·cos(2π·f2·t)], then both bounds scale by the same 0.5 factor: Enorm,max = 0.5(A1 + A2) and Enorm,min = 0.5|A1 − A2|. Full cancellation down to zero (Emin = 0) occurs only when A1 = A2.

2. How it is perceived: Roughness, pitch, and amplitude vs. loudness

How the human auditory system perceives two simultaneous tones depends on their frequency separation Δf = |f2 - f1| relative to the cochlear critical bandwidth (equivalent rectangular bandwidth, ERBN), which widens with center frequency (Moore, 2012). The perceptual regimes below are approximate and shift with carrier frequency and listening level:

  • Slow beats (Δf ≲ 10 to 15 Hz): Under suitable listening conditions when both tones fall within the same auditory filter, they are typically heard as a single fused pitch near favg whose loudness waxes and wanes smoothly at Δf times per second.
  • Flutter and auditory roughness (Δf ≈ 15 to 75 Hz at mid frequencies): As the beat rate increases, the individual loudness dips become too rapid to track one by one, producing a fast flutter or rough, buzzing texture whose peak roughness rate depends on the carrier frequency.
  • Resolved two-tone perception (Δf exceeding the critical bandwidth): When the frequency spacing becomes wide enough for the basilar membrane to resolve the two tones into separate cochlear channels, the beating sensation fades and the listener hears two distinct pitches simultaneously.

In acoustic descriptions, it is also important not to conflate three distinct quantities:

  • Signal amplitude: The instantaneous or peak mathematical excursion of the waveform (for example, normalized between -1.0 and +1.0, or sound pressure in pascals).
  • Acoustic intensity and power: For a plane progressive wave in a medium of characteristic acoustic impedance ρc, acoustic intensity I is proportional to the mean-square sound pressure (I = prms2 / (ρc)), whereas total acoustic power integrates intensity over a surface area (Hartmann, 1997).
  • Perceived loudness: A subjective psychoacoustic sensation that depends non-linearly on sound pressure level, frequency, bandwidth, and individual hearing sensitivity (ISO 226:2023; Moore, 2012).

3. Physical vs. dichotic vs. pulsed rhythmic sound

Because the word "beats" is used loosely in popular media, it helps to distinguish four different ways of creating a periodic auditory rhythm:

Method How the rhythm is generated Where modulation exists Typical perceptual character
Monaural beats Two close frequencies (f1, f2) summed in the same channel or in the air before the ear. In the physical acoustic waveform (100% envelope depth when A1 = A2). Clear, physical pulsing heard through either headphones or speakers.
Binaural beats f1 sent only to the left ear and f2 sent only to the right ear via stereo headphones. No amplitude modulation in either headphone cup; perceived centrally via binaural phase interaction. Subtle, gentle wavering inside the head; depth is much shallower than monaural or isochronic pulses (Grose et al., 2012).
Isochronic tones A single carrier tone multiplied by a periodic on/off amplitude envelope. In the physical waveform as explicit tone pulses separated by quiet intervals. Distinct, rhythmic tone pulses heard over headphones or speakers.
Alternating tones A tone or pulse that alternates smoothly or discretely between left and right stereo channels. Across the left and right channel gain envelopes over time. Spatial left–right movement rather than pitch interference.

4. Interactive browser illustration (5-second acoustic preview)

The interactive synthesizer below runs locally in your browser using the standard Web Audio API. It lets you compare how four rhythmic approaches sound using the same base frequency (211 Hz).

Important note before listening

This preview is a simple acoustic illustration, not a hearing test, diagnostic tool, or demonstration of brainwave entrainment. Perception varies across individuals, headphone designs, and room acoustics, and comparing two uncalibrated browser stimuli is not a controlled test of perceptual or clinical effectiveness. Start with your device volume set low, and use stereo headphones if you wish to hear dichotic binaural or left–right alternating separation.

Compare 4 rhythmic synthesis methods (5 s auto-stop)

Synthesized in real time with 200 ms linear fade-in/fade-out, a short 30 ms stop fade intended to reduce abrupt transitions, and conservative peak gain (0.15).

Select any button above to play a 5-second acoustic illustration.