Parametric Loudspeaker
By exploiting the fact that air is ever-so-slightly nonlinear, you can hide a conversation inside a beam of inaudible ultrasound and have it demodulate itself into sound exactly where you aim.
Projects a highly directional beam of audible sound — voice, music, or a specific audio cue — with a beam width of roughly 3–5° (compared to 30° or more for a traditional loudspeaker of the same physical size). Because the audible sound is generated progressively along the ultrasonic beam column, not at the face of the transducer, the effective acoustic source appears to float in mid-air. Listeners outside the beam hear nothing. Used in museums, retail displays, defence 'acoustic hailing' devices, and audio privacy systems.
The audio signal is amplitude-modulated onto an ultrasonic carrier, typically 40–60 kHz (wavelength ~7–8 mm in air). Because the transducer aperture (typically 0.2–0.6 m) is many wavelengths wide at the carrier frequency, the ultrasound propagates as a near-diffraction-limited beam — just as a large telescope produces a narrow optical beam. As this high-intensity column travels through air, the medium's slight nonlinearity (captured in the Westervelt equation and Berktay's far-field solution) causes the wave to distort: energy at the carrier and its modulation sidebands interacts to generate new frequency components, including difference-frequency terms that fall in the audible range. This self-demodulation is cumulative over distance and produces an apparent audio source spread along the beam axis, with no sidelobes. The audio signal appears to originate not at the transducer but out in the room, precisely where the beam terminates. Achieving low harmonic distortion requires pre-distorting the drive signal with a square-root or single-sideband preprocessing step; simple AM modulation produces substantial second-harmonic distortion in the demodulated audio.
In plain terms
Imagine shining a torch through a special kind of glass that only lets through red light. The torch beam travels in a straight line, the glass strips out everything but red, and you end up with a narrow red spotlight. A parametric speaker does something similar with sound: it fires an invisible ultrasound beam, and the air itself — like that special glass — strips out the ultrasonic carrier and leaves behind only the audible music or speech, right where the beam lands.
Also called: audio spotlight, ultrasonic directional speaker, parametric acoustic array (PAA), audio beam speaker.
Why it matters
- A parametric speaker 0.3 m across achieves narrower audio directivity at 1 kHz than a conventional loudspeaker 7 metres across — physics that is otherwise unreachable at audio wavelengths without enormous arrays.
- The demodulated audio beam has virtually no sidelobes, unlike phased loudspeaker arrays whose pattern control breaks down at low frequencies, making it suitable for directing sound to one visitor in an exhibit while adjacent visitors hear nothing.
- Because the sound is generated in air rather than at the transducer face, the beam can be reflected off a flat wall to create an apparent audio source at the reflection point, enabling behind-screen or around-corner audio effects.
- Practical commercial systems (Holosonics Audio Spotlight, SoundTube IP-Beam) have been field-deployed since the early 2000s, confirming that the technology is a production reality — though with genuine limitations in low-frequency reproduction and absolute output level.
Where it lives
- Holosonics Audio Spotlight AS-24i — commercial 60 cm flat-panel unit, ~200 Hz–20 kHz response, 5° beam divergence, used in museums and retail installations worldwide
- American Technology Corporation (ATCO) LRAD / HyperSpike — acoustic hailing devices; parametric audio at longer ranges for maritime and perimeter security
- Sennheiser AudioBeam — discontinued parametric speaker used in European museum and broadcast installations, ~40 kHz carrier
- Custom 40 kHz PZT or piezo-film arrays — research and maker community builds, widely documented; the ultrasonic carrier attenuates ~1 dB/m in air at 40 kHz, setting a practical maximum range of ~5–10 m for most array sizes
Key numbers
- Carrier frequency (typical): 40–60 kHz (Inaudible to humans; wavelength 6–8.5 mm sets the beam geometry)
- Beam divergence (Holosonics AS-24i): ~5° (Versus ≥30° for a conventional speaker of equal aperture at 1 kHz)
- Usable audio frequency range: ~200 Hz – 20 kHz (Low-frequency floor imposed by Berktay demodulation physics, not amplifier power)
- Carrier attenuation in air: ~1 dB/m at 40 kHz (Sets practical effective range to ~5–15 m for typical arrays)
- Required carrier SPL for demodulation: >110–120 dB at face (Nonlinear air interaction requires high intensity at the transducer aperture)
- THD (with root-square AM preprocessing): Substantially reduced vs. simple AM (Pompei 1998 AES Conv. Preprint 4853; exact figures are system-dependent)
Surprising facts
- The carrier frequency is completely inaudible — at 40 kHz, the ultrasound passes silently through the room. Only the audio component that 'grows' inside the beam as it travels is audible, and only to someone standing in the beam's path.
- Westervelt published the theoretical foundation for parametric arrays in the Journal of the Acoustical Society of America in 1963. The first experimental demonstration in air (not water) was not achieved until 1975, more than a decade later, because generating the necessary ultrasonic intensities in open air was technically demanding.
- The main limitation of parametric speakers is not directionality but fidelity: the Berktay demodulation process emphasises high frequencies and attenuates bass. Below roughly 200–400 Hz, practical parametric systems produce very little output — a limitation that is partly, but not fully, compensated by electronic pre-emphasis.
- Early attempts by Matsushita (Panasonic) and others to commercialise the technology in 1983 failed because the total harmonic distortion exceeded 50% — the demodulated audio was so distorted it was barely intelligible. Joseph Pompei's 1998 preprocessing insight (presented at the AES 105th Convention) reduced distortion to acceptable levels and made practical products possible.
Open questions
- Anomalous off-axis audio perceived by nearby listeners: In several field deployments, listeners clearly outside the nominal 5° beam report hearing faint audio content from parametric speakers. The effect is not fully explained by simple sidelobe models. One candidate mechanism is re-radiation of audio from reflective surfaces that intercept the ultrasonic beam; another is bone-conduction coupling when the body partially intersects the beam edge. The relative contributions have not been cleanly separated in published studies, and the perceptual boundary of the effect varies unpredictably with room geometry.
A short history
The parametric acoustic array — using the nonlinear interaction of two collinear sound beams to generate a third, difference-frequency beam — was first formalised theoretically by Peter Westervelt at JASA in 1963, building on his earlier work on scattering of sound by sound. The concept was originally developed for underwater sonar, where water's nonlinearity is exploitable at modest intensities. The first experimental demonstration in open air was achieved by Bennett and Blackstock in 1975, and the first attempt at an audible air-borne audio device was published by Yoneyama and colleagues (Matsushita / NTT group) in JASA in 1983 — but distortion was >50% and the project was abandoned. The modern era began when Joseph Pompei, then a graduate student at Northwestern University and later at MIT, recognised that pre-processing the drive signal (presented at the AES 105th Convention, Preprint 4853, 1998) could compensate for the inherent second-harmonic distortion of the demodulation process. Pompei founded Holosonics Research Labs (now Holosonics), which has manufactured the Audio Spotlight product since the early 2000s.
Sources
- Westervelt, J. Acoust. Soc. Am. 35, 535 (1963) — parametric acoustic array (foundational theory) (https://doi.org/10.1121/1.1918525)
- Yoneyama et al., J. Acoust. Soc. Am. 73, 1532 (1983) — first air-borne parametric loudspeaker demonstration (https://doi.org/10.1121/1.389414)
- Pompei, F.J., AES 105th Convention Preprint 4853 (1998) — Audio Spotlight preprocessing (https://www.holosonics.com/sounds-from-ultrasound-history)
- Holosonics Audio Spotlight — FAQ and technical specifications (https://www.holosonics.com/faq)
- Shi & Gan, Applied Acoustics — review of parametric acoustic array in air (2010) (https://convexoptimization.com/TOOLS/ReviewParametricAcoustics.pdf)