The Structure of Whale Song

Marine mammal vocalizations are among the most complex acoustic productions known in the biological world. Several lines of evidence converge: the songs are hierarchically structured (units → phrases → themes → songs); they are culturally transmitted (entire breeding populations sing the same song; songs evolve over seasons); they exhibit dialect variation between populations; sperm whale codas show combinatorial coding of meaning; and they propagate through an ocean acoustic channel — the SOFAR — that lets signals carry hundreds to thousands of kilometers. Whale song is, in an important sense, the longest-range biological communication system on Earth.

This article surveys the structure of humpback song, sperm whale codas, blue whale low-frequency calls, and right whale up-calls; the SOFAR channel and ocean acoustics; the cultural transmission and dialect-group findings; and the modern computational efforts (Project CETI, Earth Species Project) to decode these systems. It complements the Marine Communication module's full report and bridges to the Cymatics and African Drumming articles in this module.

Humpback whale song — Megaptera novaeangliae#

The vocalizations of male humpback whales on their breeding grounds are the most intensively studied animal songs in biology. Roger Payne and Scott McVay's 1971 Science paper "Songs of Humpback Whales" — based on World War II underwater hydrophone recordings — established the basic facts that have shaped the field ever since.

Hierarchical structure#

Humpback song decomposes into nested levels:

| Level | Component | Typical duration | |---|---|---| | Unit | A single sound — a moan, cry, chirp, click, or whup | 1–5 seconds | | Phrase | A sequence of 2–8 units repeated as a fixed pattern | 7–15 seconds | | Theme | A series of related phrases | 2–4 minutes | | Song | A complete sequence of themes (5–8 typical) | 10–30 minutes | | Session | Continuous singing of multiple songs | Up to 24+ hours |

Each level is a stable structural unit: phrases are repeated identically; themes always contain the same set of phrases; the song's themes appear in a fixed order. A male humpback singing on his Hawaiian, Caribbean, or Tongan breeding ground produces a song that has internal grammatical structure remarkably similar to musical composition.

Cultural transmission#

Two findings from Payne, Hal Whitehead, and Ellen Garland transformed the understanding of whale song from "fixed signal" to "cultural product":

1. All males in a breeding population sing the same song. A single male's individual song is the breeding population's collective song; Garland and others established this by recording multiple individuals and confirming the strict overlap.
2. The song evolves gradually over each season — and all singers adopt the changes simultaneously. New phrase variants appear; established phrases are dropped; theme orderings shift. The whole population tracks the evolving song. Across seasons (decades), the song is unrecognizable from a recording of decades earlier; it is never exactly repeated.

Garland et al. (2011, Current Biology) documented "song revolutions" in the Pacific: a song from one population (Australia / eastern Australian breeding ground) was wholesale adopted by an adjacent population (New Caledonia, Fiji, Tonga) in successive years, replacing the receiving population's previous song. This is horizontal cultural transmission at a scale rare in non-human animals — comparable in some respects to the transmission of human dialect or fashion.

Musical-style features#

Humpback song has been analyzed by musicologists for several decades. Findings include:

• Theme structure (ABCDE...) — themes appear in fixed order, not at random.
• Rhyming patterns — Guinee and Payne (1988) identified phrase pairs ending in similar acoustic units, structurally analogous to the rhyming used in human poetry as a memory aid.
• Tempo and meter — phrases tend to maintain consistent durations within a song, suggesting an internal "tempo."
• Frequency range — most energy in 100–4,000 Hz, with components extending from 30 Hz up to 8,000 Hz; pitch and register are systematically used.

Whether humpback song is "music" in any human-comparable sense is a contested but interesting question. The structural features that make human music recognizable as such — hierarchy, repetition, evolution, culture — are unmistakably present in humpback song.

Sperm whale codas — Physeter macrocephalus#

The sperm whale produces patterned sequences of broadband clicks called codas. Each coda is a short (often 0.5–2 second) sequence with a specific click-timing pattern. Unlike humpback song, sperm whale codas are short, click-based rather than tonal, and used in social communication rather than (apparently) breeding display.

Combinatorial coding — the 2024 Project CETI findings#

The 2024 Nature paper by Andreas, Beguš, and the Project CETI team analyzed Eastern Caribbean sperm whale codas and identified:

• At least 142 distinct coda types in the corpus, characterized by click count, timing, and modulation.
• A combinatorial coding system: meaning is constructed by combining basic elements (rather than each coda being an atomic unit). The team identified what they called rubato (timing variation), rhythm, tempo, and ornamentation as features that vary across multiple distinct dimensions and combine to produce the full coda repertoire.
• Context-dependent variation: codas change based on social situation. The same animal produces different coda variants depending on which other individuals are present.
• Dialect groups — distinct clans of sperm whales in the same ocean basin produce distinct coda repertoires; clan membership is more important than geography.

The combinatorial structure is significant. If each coda were a holistic signal, the maximum information content would be ~7 bits (log₂ 142). Combinatorial coding lets the same set of elements encode much more information through combinations — closer to how human phonology builds many words from few phonemes. The 2024 paper does not claim the system is "language" in a strong technical sense, but it does claim that the coda system has the structural prerequisites that linguists generally take to be present in human language.

Clan structure#

Hal Whitehead and Luke Rendell's Cultures in the Toothed Whales of the Sea (Whitehead and Rendell, 2014) documented vocal clans: groups of sperm whales that share coda repertoires across hundreds of kilometers. Five major Pacific clans were initially identified (Whitehead 2003); subsequent work identified additional clans in the Atlantic, Indian, and Mediterranean. Clan membership is matrilineal — calves inherit their mother's clan repertoire.

The clan system is remarkable: the cultural unit is not the geographic population (as it is for humpbacks) but the social-genealogical lineage. Two clans can occupy overlapping waters and use distinct coda repertoires; the system is functionally equivalent to two human cultures sharing a city.

Blue whale calls — Balaenoptera musculus#

The blue whale produces some of the lowest-frequency biological vocalizations on Earth — calls centered around 10–25 Hz, below the lower bound of human hearing (~20 Hz). Blue whale calls are loud (up to ~190 dB at 1 meter), simple in structure (typically pulsed tones lasting 5–15 seconds), and propagate extraordinarily far through ocean water.

The North Atlantic, North Pacific, and Antarctic blue whale populations have distinct call types — call variation that has been used to distinguish separate breeding populations even when individuals visually mix. McDonald, Hildebrand, and Mesnick (2009) documented a gradual decrease in pitch over decades of blue whale calls — a slow, hemispheric-scale acoustic drift whose cause is not understood (proposed: increasing population density, ocean noise pollution, social-cultural drift, individual aging effects).

Right whale up-calls — Eubalaena glacialis#

The North Atlantic right whale produces a characteristic up-call — a frequency-rising tonal call lasting 1–2 seconds and used widely between individuals across hundreds of kilometers. The up-call is the principal contact-and-cohesion signal for the species, which is critically endangered (~ 360 individuals as of 2023). Acoustic monitoring of right whale up-calls in the Gulf of Maine and Bay of Fundy is the principal tool used to direct shipping-route closures and reduce ship-strike mortality.

The SOFAR channel — long-range ocean acoustics#

Whales' acoustic range depends on the SOFAR channel (Sound Fixing and Ranging) — a horizontal layer of ocean water (typically 600–1200 m depth, varying by ocean) where the speed of sound reaches a minimum, creating a waveguide that traps low-frequency sound and propagates it with minimal attenuation.

The channel was discovered in 1944 by Maurice Ewing and J. Lamar Worzel for the U.S. Navy: shells exploded in the channel were detected from over 5000 km away. The same channel that carries military signals carries blue whale calls; the calls have been recorded across entire ocean basins.

For humpbacks, sperm whales, and right whales — which sing or call from shallower depths — the channel still helps but is less efficient. Their effective ranges are shorter (perhaps 20–200 km depending on species and conditions), but still much greater than terrestrial-mammal vocal range.

Cetacean acoustic anatomy#

The mechanisms by which whales produce these sounds are themselves remarkable:

Toothed whales (Odontoceti) — clicks via phonic lips#

Sperm whales, dolphins, killer whales, and other toothed whales produce sound through phonic lips (also called museau de singe) — paired structures in the nasal passage that vibrate as air is pushed through. The vibration is amplified and focused through the spermaceti organ and the melon (a fatty acoustic-lens structure in the forehead) into a directional beam projected forward.

The sperm whale's spermaceti organ — once hunted to near-extinction for the high-quality wax it contains — is now understood to be an acoustic-focusing apparatus. The 1820 sinking of the whaling ship Essex by a sperm whale (the inspiration for Moby Dick) involved an animal whose head contains the largest acoustic-imaging system in the natural world.

Baleen whales (Mysticeti) — vocalization via laryngeal U-fold#

Humpbacks, blue whales, right whales, and other baleen whales produce sound through a U-shaped laryngeal fold and the inflation/deflation of the laryngeal sac. The mechanism is closer to mammalian larynx-driven vocalization but adapted to underwater acoustics.

The recent paper by Coiro Reidenberg and Damien Reigner (2024, Nature) described in detail the baleen whale laryngeal anatomy and demonstrated experimentally how the structures produce the observed call frequencies — closing a long-standing gap in cetacean acoustic biology.

Modern computational efforts#

Project CETI (Cetacean Translation Initiative)#

Founded 2020 by David Gruber, Shane Gero, and a multi-disciplinary team. CETI's mission is to record, decode, and (eventually) communicate with sperm whales using machine learning. Field site: Dominica. Methodology: dense hydrophone arrays, drone-based bioacoustic tagging, and large-scale modern ML applied to coda corpora. The 2024 Nature paper (combinatorial coding findings) is the project's first major output.

CETI's research design is deliberately patient: the team has explicitly avoided premature claims of "translation," focusing instead on structural mapping of the coda system and its social context. The Foundation is privately funded; its scientific advisory includes Robert Wood, Patrick Hof, Daniela Rus, Pratyusha Sharma, and Roger Payne (until his death in 2023).

Earth Species Project#

Founded 2017 by Aza Raskin and Britt Selvitelle, ESP applies machine learning across multiple animal communication systems (whales, dolphins, primates, birds). ESP has released open-source bioacoustic ML models, public datasets, and is developing foundation models for animal communication.

Other research programs#

• Sapphire Project — humpback song ML analysis, Cornell Lab of Ornithology.
• MARS / NOAA acoustic monitoring — long-term marine acoustic data infrastructure.
• Wildlife Conservation Society Ocean Giants Program — long-term humpback song study in the North Atlantic.
• Sea Mammal Research Unit (St Andrews) — UK-based marine bioacoustic and behavior work.

The 2020s are the most productive decade in marine bioacoustics history; computational power, hydrophone networks, and deep learning have transformed what is empirically possible.

Cultural transmission as the underlying frame#

What unites the humpback, sperm whale, blue whale, and other cetacean systems is cultural transmission. Each population's vocal output is not simply a fixed species-typical signal; it is a learned, transmitted, and dynamically evolving cultural product. This places cetaceans in a small group of taxa with documented vocal culture: humans, songbirds (especially zebra finch and a few cosmopolitans like starlings and parrots), bats (some Microchiroptera), and a few primates (recently documented cultural call traditions in some macaque populations).

The implication: the diversity of whale song is not just acoustic diversity. It is cultural diversity — and lost populations carry with them lost cultures, lost traditions, lost songs that may never be recovered.

Connection to this knowledge base#

• The Cymatics article documents the visualization of structured sound; the John Stuart Reid CymaScope has been used specifically to analyze cetacean vocalizations.
• The African Drumming and Ritual Rhythm article and the Sufi Dhikr and Zikr article cover human structured-sound traditions that share certain communicative properties with whale song (cultural transmission, ritual context, hierarchical structure).
• The Marine Communication module is the deeper home for marine-mammal cognition and communication research; this article focuses on the acoustic-structural properties.
• The Frequency Reference doc provides numerical reference for whale vocalization frequency ranges.

Sources#

• Andreas, Jacob, Gašper Beguš, et al. "Context-dependent combinatoriality in sperm whale vocalizations." Nature 634 (2024).
• Cerchio, Salvatore, Paul O. Andrianantenaina, Anjara Saloma, Daniel Rasoloarijao, Howard Rosenbaum, et al. "Acoustic monitoring of humpback whale song revolutions in the southwest Indian Ocean." Endangered Species Research (2018).
• Coiro Reidenberg, Joy, and Damien Reigner. "The baleen whale laryngeal U-fold." Nature (2024).
• Garland, Ellen, Anne Goldizen, Melinda Rekdahl, Rochelle Constantine, Claire Garrigue, Nan Daeschler Hauser, M. Michael Poole, Jooke Robbins, and Michael Noad. "Dynamic Horizontal Cultural Transmission of Humpback Whale Song at the Ocean Basin Scale." Current Biology 21 (2011).
• Guinee, Linda N., and Katherine Payne. "Rhyme-like Repetitions in Songs of Humpback Whales." Ethology 79 (1988).
• McDonald, Mark A., John A. Hildebrand, and Sue Mesnick. "Worldwide decline in tonal frequencies of blue whale songs." Endangered Species Research 9 (2009).
• Payne, Roger S., and Scott McVay. "Songs of Humpback Whales." Science 173 (1971): 585–597.
• Rendell, Luke, and Hal Whitehead. The Cultural Lives of Whales and Dolphins. University of Chicago Press, 2014.
• Sharma, Pratyusha, Jacob Andreas, Gašper Beguš, et al. (Project CETI publications, 2022–2024).
• Whitehead, Hal. Sperm Whales: Social Evolution in the Ocean. University of Chicago Press, 2003.
• Whitehead, Hal, and Luke Rendell. "Movements, communication, and culture in sperm whales." Annual Review of Ecology, Evolution, and Systematics 50 (2019).