What Humpback Whales Are Communicating Through Their Songs and Who Is Listening in the Ocean

A single humpback whale singing in the North Pacific can produce a sound sequence lasting up to 30 minutes, then repeat it almost exactly. Over weeks, the song changes, new phrases inserted, old ones dropped, and every male in the same ocean basin makes the same update, in the same order, as if following a score that nobody wrote down. This is not metaphor. Roger Payne and Scott McVay documented this coordinated evolution of humpback songs in their 1971 paper in Science, and the pattern has held in every ocean population studied since. The songs are not random. They are structured into units, phrases, and themes, a hierarchy that linguists use to describe human language, borrowed here because nothing else fits.

The biological mechanism behind the sound is unusual. Humpbacks have no vocal cords. They push air through a specialised laryngeal structure and a system of inflatable sacs, recycling the same air without exhaling. A whale can sing continuously for hours without surfacing. The frequencies range from about 20 Hz, below what most humans can hear without equipment, up to 24,000 Hz, spanning a range that crosses the hearing thresholds of fish, dolphins, and other baleen whales entirely.

The audience is layered. Other male humpbacks in the same region hear the song and, researchers believe, use it to gauge competitive fitness, the complexity and duration of a song may signal stamina and health. Females listen too, though the exact role of song in mate selection remains an open question. A 2022 study by researchers at the University of Queensland found that female humpbacks spent more time near males producing longer, more varied song sequences, which suggests preference, but the causal chain is not yet closed.

Beyond other humpbacks, the ocean is full of listeners that scientists did not initially consider. Sperm whales, orcas, and fin whales all share acoustic frequency ranges that overlap with humpback song. Whether they parse meaning from it or simply register it as environmental noise is unknown. What is known is that the sound travels. In the deep sound channel, a layer of ocean where temperature and pressure combine to trap and propagate low-frequency sound, humpback songs can travel up to 10,000 kilometres without significant loss. A whale singing off the Andaman Islands could, in principle, be heard by another whale near Madagascar.

The honest answer is that full semantic decoding has not happened. But the structural evidence points toward several functions running simultaneously. Song appears to serve mate attraction, male-to-male competition signalling, and possibly group coordination during migration, three functions that are not mutually exclusive and may be layered into the same acoustic sequence depending on context. The cultural transmission angle is the most striking: new song elements spread across ocean basins from west to east, documented in the Pacific by Michael Noad's team at the University of Queensland. A new phrase originating in Australian waters was picked up by whales in French Polynesia within two years. This is not genetic. It is learned behaviour spreading through a population, acoustic culture.

Google and the nonprofit Wild Me launched Project Cetacean Translation Initiative (CETI) to apply machine learning to sperm whale click sequences, but parallel efforts are now targeting humpback song. The NOAA Fisheries acoustic monitoring network has accumulated decades of continuous underwater recordings from hydrophone arrays across the Pacific and Atlantic. AI models trained on these archives are identifying micro-variations in phrase structure that human analysts missed, not because human ears are inadequate, but because the dataset is simply too large for manual review. What the models are finding is that individual whales have consistent acoustic signatures within the shared song: a kind of vocal fingerprint layered inside a communal text.

The military dimension is less discussed but real. Naval sonar systems, particularly low-frequency active sonar used by the US Navy, operate in frequency ranges that directly overlap with humpback communication bands. The overlap is not incidental. Several strandings of cetaceans near naval exercises have been linked to acoustic disruption, and a 2003 report by the International Whaling Commission flagged mid-frequency sonar as a contributing factor in documented mass strandings. The ocean's acoustic environment, in which humpback song evolved over millions of years, now carries industrial shipping noise, seismic survey pulses, and military sonar, all of which the whales cannot filter out.

The question humpback song keeps forcing back onto researchers is the one they are most reluctant to answer directly: does this constitute language? The structural criteria, discrete units, hierarchical organisation, combinatorial variation, cultural transmission, are present. What is missing is demonstrated referential meaning: evidence that a specific phrase points to a specific thing in the world. That gap may be a genuine cognitive difference, or it may be a measurement problem. Humans have been studying humpback song for about 50 years. The whales have been refining it for far longer.

The song, the silence around it, and the new noise cutting through both are the same story told from three directions: a species that evolved one of the most complex communication systems on the planet, a set of technologies finally sensitive enough to begin reading it, and a set of other technologies loud enough to drown it out before the reading is done.