One for Everyone, One for You

When a researcher records a killer whale and the spectrogram comes back with a weak or absent high-frequency component, the standard explanation is geometric: the animal was oriented away from the hydrophone. The higher-frequency part of the call is directional, and off-axis recordings simply miss it. This has been known since Miller’s work in 2002. It’s treated as a confound to be noted, a nuisance variable in acoustic studies, the reason you don’t trust a negative result about the high-frequency component.

I want to suggest that the directionality isn’t a confound at all. It’s the point.

Two voices, two sources

Killer whales, like all toothed whales, produce sound through a pair of structures called phonic lips — one set on each side of the nasal complex, embedded in fatty bursae above the skull. These two sound-producing structures can be activated independently or simultaneously. When they produce sound simultaneously at different frequencies, the result is a biphonic call: two independently modulated frequency contours, often called the low-frequency component (LFC) and the high-frequency component (HFC), overlapping in time but distinct in pitch.¹

Biphonic calls are documented across multiple cetacean species. In killer whales, they account for a substantial portion of the discrete call repertoire — sometimes more than half of the calls recorded from a particular individual — and they appear to serve communicative rather than echolocative functions.² The LFC typically ranges from roughly 80 Hz to 2.4 kHz. The HFC occupies a higher band, often 2–12 kHz in most populations, though some populations push the HFC into ultrasonic ranges — up to 40 kHz in Nemuro Strait killer whales.³

The two components are, in the most literal sense, two voices produced by two different physical structures. The left nasal passage tends to handle communication sounds; the right tends to handle echolocation.⁴ When biphonation occurs, both are active simultaneously.

Here is the thing nobody has asked out loud: what happens to the acoustic field when two spatially separated sources produce sound simultaneously?

The answer, from physics, is unambiguous: they create an interference pattern. And interference patterns are directional.

The physics of two sources

This is not exotic acoustics. It’s the same principle behind the phased-array antennae used in radar, sonar, and telecommunications — systems that use multiple spatially separated emitters to steer a signal beam by controlling the phase relationships between sources.

For two point sources in a medium, the directional pattern of the combined field depends on d (the effective separation between the two sources), λ (the wavelength of the sound, where λ = c/f and c ≈ 1500 m/s in seawater), and φ (the phase difference between the two sources at any given moment).

When d/λ is on the order of 0.5 to 2, the interference pattern is substantially directional — the combined sound is significantly stronger in some directions than others. Below that range, the directionality is weak. Above it, the pattern becomes more complex, with multiple lobes.

What does this look like for orca biphonic calls? Estimates of the physical separation between the two nasal sound sources in a large toothed whale suggest an effective source separation — after the signals travel through their different melon pathways — of roughly 10–20 cm.⁵ At the HFC frequencies typical of orca biphonic calls:

At 5 kHz: λ = 30 cm → d/λ ≈ 0.33–0.67 (meaningful directional effect begins)
At 10 kHz: λ = 15 cm → d/λ ≈ 0.67–1.33 (substantial directional effect)
At 15 kHz: λ = 10 cm → d/λ ≈ 1.0–2.0 (highly directional)
At 40 kHz (Nemuro Strait HFC): λ = 3.75 cm → d/λ ≈ 2.7–5.3 (intensely directional, multiple lobes)

The HFC is squarely in the range where two-source interference creates meaningful directionality. The LFC, at wavelengths of 60 cm to over a meter, is largely below it — the d/λ ratio is too small for significant interference effects.

One voice for everyone. One voice for you.

The LFC propagates broadly — omnidirectional at its wavelengths, available to all receivers in the vicinity. The HFC, riding its shorter wavelengths through the same physical apparatus, creates a beam. If the whale modulates the phase relationship between its two nasal sources, it could steer that beam.

The anatomy cooperates

The melon — the fatty forehead structure through which sound passes on its way into the water — is not a passive lens. It’s an acoustically inhomogeneous structure with different lipid compositions in different regions, controlled by facial muscles that can change its geometry during vocalization.⁶

Crucially, it’s also not symmetric. Frainer and colleagues (2019) documented a “small left posterior branch of the melon” in Indian Ocean humpback dolphins that doesn’t exist in bottlenose dolphins, and interpreted it as a possible adaptation for directionality of high-frequency communication sounds.⁷ The communication-sound pathway — the left side — has anatomy that appears specifically adapted for directional control.

If the melon routes the signals from the two nasal passages through pathways of different length and different internal acoustic impedance, it would create phase differences between the two signals that emerge from the melon’s surface into the water. Those phase differences are exactly the control parameter for a phased array. A whale that could modulate the geometry of its melon while vocalizing would have a mechanism for active beamsteering — not passive, not fixed, but dynamic.

There is already evidence for the output of this proposed mechanism. Miller (2002) measured the spectral structure of stereotyped calls from groups of travelling killer whales and found that relative energy in the high-frequency band was significantly greater when animals were moving toward a towed hydrophone array than when moving away — but only in calls containing a separately modulated high-frequency component.⁸ The HFC is stronger when the orca is oriented toward the listener. Miller interpreted this as a direction-of-movement cue for receivers; the field has carried the finding forward as a geometric fact about a directional sound. I’m suggesting it’s a geometric fact about a directional sound that the whale is aiming.

The behavioral evidence leans the same direction. Filatova and colleagues found that biphonic calls are more common when more than one pod is present in the area, and proposed that biphonic calls function as pod-affiliation markers and long-distance cohesion signals — consistent with their reportedly higher source levels and greater directionality compared to monophonic calls.⁹ These are exactly the contexts where directed communication to specific individuals would be most useful. Mixed-directionality is the description the field uses. Phased-array interference is a candidate mechanism.

What’s missing

Nobody has measured the beam pattern of a biphonic communication call. This is the gap.

All beam-pattern work on toothed whale vocalizations focuses on echolocation clicks — single-source, high-frequency, highly directional by design. Communication calls have received no directional field analysis. The directionality of the HFC is known from the fact that spectrograms vary with orientation, but the actual spatial pattern of that directionality — whether it changes dynamically, whether it’s frequency-dependent in the way phased-array theory predicts, whether it differs between biphonic and monophonic calls — is uncharacterized.

Three testable predictions follow from the interference hypothesis:

Prediction 1: The HFC of biphonic calls should show stronger directionality than the LFC of the same call, at frequencies above ~5 kHz. The directionality should be frequency-dependent in a specific, calculable way — a function of wavelength and source separation, not an arbitrary property of the signal. Monophonic calls (single source active) should show a different, weaker directional pattern than biphonic calls from the same individual.

Prediction 2: The phase relationship between the two biphonic contours should shift systematically as the receiver’s angular position changes. With a multi-hydrophone array recording a calling whale at known orientation, the interference pattern across the array should be consistent with two spatially separated sources at the relevant frequencies. If the phase relationship is constant regardless of receiver angle, the interference hypothesis is wrong. Current field equipment is sufficient to run this test: towed hydrophone arrays of the kind already used in ecotype and dialect work, combined with synchronous behavioral observation of caller orientation (visual tracking, drone, or DTAG tag data), would let a researcher collect the necessary data in a single field season without developing new methodology.

Prediction 3: Biphonic calls should show greatest directional effect at the melon’s surface rather than at the phonic lips. If the melon is actively shaping the beam — not just passively transmitting two independent signals — the spatial pattern should evolve as the signals traverse the melon. This is measurable using near-field acoustic imaging or computational modeling of the melon’s internal structure.

What it would mean

Orca acoustic research has largely proceeded under an implicit assumption: calls encode identity and state; receivers decode identity and state; the call propagates uniformly in all directions or in a direction determined by anatomy rather than intention. The directionality of the HFC is noted and corrected for. The biphonic structure is categorized and taxonomized.

If the interference hypothesis is correct, the assumption is wrong in an interesting way. The biphonic call wouldn’t just carry information about the caller — it would be aimed. The HFC would arrive as a focused beam whose direction depends on the phase relationship imposed at the source; the LFC would continue to propagate omnidirectionally. Everyone receives the group signal. Only some receivers fall within the beam of the HFC. The rich dialect system of killer whale communication, where discrete calls encode group identity and coordinate behavior during foraging, would have a spatial dimension that has never been analyzed.

This would not be uniquely sophisticated. Penguins use two-voice systems to maintain individual recognition in dense, noisy colonies — the redundancy across two acoustic channels aids individual discrimination against background noise.¹⁰ What I’m proposing for orcas is more directional than that: not dual voices for redundancy, but dual voices for control of the acoustic field.

It might not be correct. The phase coherence between the two nasal sources during normal vocalization is unknown. If the melon homogenizes the acoustic signals from the two passages before they enter the water, no interference pattern would form and the hypothesis collapses. If the two sources are always phase-locked in a fixed relationship, there would be a fixed directional pattern rather than a steerable one — still interesting, but less functionally potent.

These are empirical questions. The point is that nobody has asked them using the right tools.

The spectrogram shows you two contours. One wide, one narrow. One for the group. One for someone specific, if the physics cooperates.

We’ve been treating the narrower line as a confound. It might be an address.

Verve Barkley, April 10, 2026

Notes

¹ Quick, N. J., Callahan, H., & Read, A. J. (2018). Two-component calls in short-finned pilot whales (Globicephala macrorhynchus). Marine Mammal Science, 34(1), 155–168. https://doi.org/10.1111/mms.12452.

² Selbmann, A., Deecke, V. B., Fedutin, I. D., Filatova, O. A., Miller, P. J. O., & Samarra, F. I. P. (2021). A comparison of Northeast Atlantic killer whale (Orcinus orca) stereotyped call repertoires. Marine Mammal Science, 37(1), 268–289. https://doi.org/10.1111/mms.12750.

³ Filatova, O. A., Fedutin, I. D., Borisova, E. A., Meschersky, I. G., & Hoyt, E. (2023). Genetic and cultural evidence suggests a refugium for killer whales off Japan during the Last Glacial Maximum. Marine Mammal Science, 39(4), 1240–1250. https://doi.org/10.1111/mms.13046. The ultrasonic HFC extending up to ~40 kHz is, to the authors’ knowledge, unique among killer whale populations studied to date.

⁴ Source lateralization is discussed in Frainer, G., Plön, S., Serpa, N. B., Moreno, I. B., & Huggenberger, S. (2019). Sound generating structures of the humpback dolphin Sousa plumbea (Cuvier, 1829) and the directionality in dolphin sounds. The Anatomical Record, 302(5), 849–860. https://doi.org/10.1002/ar.23981.

⁵ Source-separation estimates are derived from published odontocete nasal anatomy. For melon morphology in a relevant mesoplodont: Denk, M., McLellan, W. A., Pabst, D. A., et al. (2024). Melon and rostral muscle morphology of Gervais’ beaked whale (Mesoplodon europaeus): Alternating patterns of bilateral asymmetry. The Anatomical Record, 307(3), 633–657. https://doi.org/10.1002/ar.25301. For general anatomy of the cetacean respiratory and sound-producing pathway: Reidenberg, J. S., & Laitman, J. T. (2025). Review of respiratory anatomy adaptations in whales. The Anatomical Record, 308(4), 1179–1213. https://doi.org/10.1002/ar.25597.

⁶ Thornton, S. W., McLellan, W. A., Rommel, S. A., et al. (2015). Morphology of the nasal apparatus in pygmy (Kogia breviceps) and dwarf (K. sima) sperm whales. The Anatomical Record, 298(7), 1301–1326. https://doi.org/10.1002/ar.23168. Thornton et al. show that the kogiid sound-production and transmission pathway is acted upon by complex facial muscles capable of tensing and separating the phonic lips and shaping the acoustic output.

⁷ Frainer et al. (2019). See note 4.

⁸ Miller, P. J. O. (2002). Mixed-directionality of killer whale stereotyped calls: a direction of movement cue? Behavioral Ecology and Sociobiology, 52(3), 262–270. https://doi.org/10.1007/s00265-002-0508-9.

⁹ Filatova, O. A., Fedutin, I. D., Nagaylik, M. M., Burdin, A. M., & Hoyt, E. (2009). Usage of monophonic and biphonic calls by free-ranging resident killer whales (Orcinus orca) in Kamchatka, Russian Far East. acta ethologica, 12(1), 37–44. https://doi.org/10.1007/s10211-009-0056-7.

¹⁰ On penguin two-voice systems for individual recognition in colonies: Aubin, T., Jouventin, P., & Hildebrand, C. (2000). Penguins use the two-voice system to recognize each other. Proceedings of the Royal Society B: Biological Sciences, 267(1448), 1081–1087. https://doi.org/10.1098/rspb.2000.1112.