Anyone who enjoys spending time in restaurants, bars, or at concerts knows the phenomenon: the louder the environment, the more we have to raise our voices to be understood by the person opposite us. And we humans aren’t the only ones. The so-called Lombard effect—the adjustment of communication volume to ambient noise—is also widespread in the animal kingdom, including among birds, monkeys and marine mammals.
While we humans usually come away from a loud evening without major side effects—aside from a scratchy voice, perhaps—ambient noise poses a serious risk to dolphins and whales. Underwater, it interferes with the animals’ acoustic communication, thereby disrupting vital functions such as orientation, foraging, mate-seeking, and predator avoidance.
An effective protective strategy involves determining the animals’ location in real time and aligning human activities with their movements. On the northern Pacific coast, this is achieved using the AI-powered acoustic detection system OrcaHello. It operates around the clock to save the remaining population of Southern Resident orcas from extinction.
How AI listens to the orcas of the Salish Sea
The Spanish research project CIRCE is collecting data on porpoises on the Iberian Peninsula. The photographs in this article were taken by the project leader, Dr Philippe Verborgh. “AI-powered acoustic detection systems represent a major step forward because they filter out relevant signals from continuous hydrophone data streams in real time,” explains Dr Joseph Schnitzler from the Institute for Terrestrial and Aquatic Wildlife Research at the University of Veterinary Medicine Hannover (TiHo) to RESET. “OrcaHello, for instance, listens in around the clock at several hydrophone locations and alerts conservation groups and port authorities whenever the calls of Southern Resident orcas are detected. For the first time, this turns passive monitoring into an operational early warning system.”
The Southern Resident orcas make their home in the Salish Sea, between the US state of Washington and the Canadian province of British Columbia. With only 76 animals remaining, the population’s survival is under immediate threat. The combination of noise, food scarcity, and toxins in the water makes the situation particularly grave. “Southern Resident killer whales are in a situation where their entire habitat doesn’t offer enough quality to support their survival,” says Dr David Bain. The biologist and Vice President of the Orca Conservancy has been researching these animals for more than 40 years.
The OrcaHello system emerged from a hackathon in 2019. Today, it is run as an open-source project and has been monitoring the animals’ movements since 2020 in collaboration with the Seattle-based Orca Conservancy. To achieve this, seven underwater microphones were installed on the seabed along the coast, which continuously record ambient noise. OrcaHello’s AI listens in, detects passing whales, and triggers real-time alerts.
The technology has a decisive advantage over visual observation: “Such systems enormously increase temporal coverage, including at night, during bad weather, and in remote areas,” explains Dr Schnitzler from the TiHo. This was impressively demonstrated in September 2025. OrcaHello recorded the calls of a newborn calf at one o’clock in the morning—hours before it was spotted by humans for the first time the following morning.
Still waters are a thing of the past
But why is such a system necessary in the first place? When we humans dive into the depths of the oceans, it seems peaceful at first. Not so for the wildlife, however: underwater, sound travels about 4.5 times faster than in air. This enables whales to communicate with one another over thousands of kilometres. But anthropogenic, or human-caused, noise also bridges these distances. And this noise is constantly increasing.
Widespread sources of noise include, alongside shipping traffic, the use of sonar, the operation of offshore installations, and seismic activities during oil and gas exploration. The latter are particularly dangerous for whales. Short, artificially generated sound pulses, known as airguns, are used to search for mineral deposits on the seabed. These pulses are up to a thousand times louder than normal shipping traffic and can cause permanent damage to the animals’ hearing.
Whales are already calling as loudly as they can
A recently published study has provided new insights into the Lombard effect in pilot whales in the Strait of Gibraltar. With around 120,000 ships passing through each year, it is one of the busiest shipping routes in the world. The team led by Milou Hegeman and Frants H. Jensen, behavioural ecologists at Aarhus University in Denmark, investigated the animals’ response to the background noise. They found that the whales reach their physical limits when trying to compensate for ambient noise and are only able to partially offset it. Previous studies on other whale species have yielded similar results.
“These animals rely almost exclusively on sound to navigate their social world,” explains Hegeman. “If persistent noise reduces the range of their calls, individuals may have to call more frequently, move closer together or spend significantly more time simply maintaining contact with the group. All these responses incur both energetic and ecological costs. Even if communication is not completely blocked, chronic noise reduces efficiency and permanently alters the animals’ time budgets.”
Southern Resident orcas also adjust their vocal volume to suit their surroundings. Researchers have found that the time they need to spend foraging increases as ambient noise levels rise. Meanwhile, the salmon stocks on which the animals rely for food are already scarce. If the whales cannot find enough food, toxins accumulate in their cell membranes, where they cause health problems and developmental abnormalities in unborn calves. The Orca Conservancy is therefore working not only on monitoring with OrcaHello, but also on the restoration of salmon habitats, emergency response plans for oil spills and awareness programmes for the shipping industry.
From technology to animal welfare
The Spanish research project CIRCE is collecting data on porpoises on the Iberian Peninsula. If the OrcaHello model triggers an alert, the recording first goes to human moderators who manually confirm or dismiss it. “The passage can either be confirmed by visual sightings or assumed because the whales were detected visually or acoustically on one side of the hydrophone and then the other,” says Dr Bain. Once the animals have been located, it is crucial that a designated chain of response takes effect immediately. “The alert must go to the right places and trigger concrete measures: speed reductions, route adjustments, or pausing noise-intensive work,” emphasises Dr Schnitzler from the TiHo.
In the Salish Sea, port authorities, construction sites, ferry operators and conservation agencies are informed directly as soon as a whale sighting is confirmed. This is having an impact. “Washington State Ferries has adapted its practices based on OrcaHello data to better protect the whales,” says Dr Bain. Protection against coastal construction noise, in particular, has been improved.
Importance of AI in wildlife conservation is growing
OrcaHello is currently tailored to the orcas of the Salish Sea. However, the use of artificial intelligence detection in species conservation is expanding. “The good news is that it is becoming increasingly easier to implement AI solutions for biodiversity monitoring,” explains Dr Jensen from Aarhus University in Denmark. “Commercial systems are maturing, and there are several cost-effective platforms for audio recording and edge AI.”
In addition to the pilot whale study, his laboratory also investigated the use of AI to detect North Atlantic right whales. The relevance of such systems is growing, not least because the expansion of offshore wind power in many regions overlaps with the migration routes of endangered whale species. Acoustic monitoring enables targeted interventions that help to counteract species loss.
Training the models remains a challenge. “A system that works well in the Salish Sea is not automatically transferable to the Mediterranean or the North Sea,” Dr Schnitzler points out. Anyone wishing to develop a model for the Strait of Gibraltar needs data from precisely that acoustic environment. The fact that marine species are shifting their ranges and the timing of their migrations due to climate change complicates this endeavour.
Furthermore, not all whale species are equally easy to detect acoustically. “A non-detection does not automatically mean that there are no whales in the area,” Dr Schnitzler points out. Toothed whales such as orcas, dolphins and porpoises produce clicks and calls continuously, thereby providing reliably recognisable signals. Baleen whales, on the other hand, call less frequently and more irregularly.
According to the expert, for systems such as OrcaHello to scale up, there is a need for greater standardisation, long-term funding and a legal framework. The EU is currently pursuing an approach whereby no more than 20 percent of a target species’ habitat may exceed a biologically relevant noise threshold. However, this threshold has yet to be determined for many species and regions.
WhaleSpotter: AI protects grey whales in San Francisco Bay from collisions with ships
Climate change is altering the food chains in the Arctic, causing more and more grey whales to head for San Francisco Bay in search of food. However, they are directly endangered here by heavy shipping traffic. In 2025, 21 dead grey whales were found in San Francisco Bay – the highest figure in 25 years – 40 percent of which died as a result of collisions with ships.
The AI detection system WhaleSpotter is designed to protect the animals from collisions: the system locates the animals from a distance of up to four kilometres and alerts ships in real time so that they can alter their course. Unlike OrcaHello, which relies on acoustic monitoring, WhaleSpotter uses data from thermal cameras to detect whales.
Drive quieter, go further
AI detection systems track whales and enable protective measures to be taken. However, they do not make the oceans any quieter. The EU research project SATURN investigated which technical measures are available to reduce the impact of ship noise at source. The results show that effective noise protection is also a matter of ship design and operating practices. Optimised propellers, more aerodynamic hulls and better-insulated engine rooms lower noise levels.
However, the greatest immediate effect comes from a simple measure: sailing more slowly. “A steady, reduced speed can often reduce both noise and energy consumption at the same time,” explains Dr Schnitzler, who was involved in SATURN.
Ultimately, the same applies to the oceans as it does on land. To save biodiversity, we humans may simply need to slow down a gear. The quieter it becomes in the water, the better the whales can recover.
