OrcaHello AI System Provides Real-Time Detection of Endangered Southern Resident Orcas to Reduce Vessel and Construction Disturbance

OrcaHello, an AI-driven acoustic detection system, is monitoring southern resident orcas in real time across the Puget Sound area, helping pause coastal construction and redirect vessel traffic as the endangered animals pass through. With only 76 individuals of this subspecies remaining as of December 2025, the system represents one of the most operationally connected applications of marine AI to active conservation management, translating underwater acoustic data into timely alerts that government agencies, port operators, and fisheries managers can act on.

 

Technology Architecture and Detection Capability

 

OrcaHello builds on Orcasound, an open-source network of hydrophones installed at seven locations throughout the Puget Sound area off Washington state, which livestream underwater audio continuously. The machine-learning model developed by the OrcaHello team listens to this audio feed 24 hours a day to detect calls specifically associated with southern resident orcas, designed to minimise false positives while capturing genuine orca vocalisations with high confidence. Once a potential detection is flagged, expert moderators review the alert before it is confirmed or rejected, adding contextual information such as call type or pod identification before alerts are distributed to subscribers. The human review layer is important because it ensures that the alert system maintains scientific credibility, which is the foundation of its ability to prompt real-world operational responses.

 

Operational Impact on Construction and Vessel Management

 

The system has generated actionable alerts on more than 150 days since its inception, with tangible consequences for how coastal construction and vessel operations are managed. When OrcaHello detects orca calls near the Port of Seattle during pile-driving operations, experts contact the port to pause the noisy construction while the animals are in the area. The system also provides advance warning, sometimes detecting orca presence hours before they reach a specific construction site, allowing construction crews to reschedule to less acoustically disruptive activities during the period when orcas are expected to be in the vicinity. Chief scientist David Bain of Orca Conservancy has noted this predictive capability enables the construction crew to be informed that the orcas are likely to arrive around a specific time, making forward planning possible rather than reactive pause.

 

Species Context and Conservation Urgency

 

Southern resident orcas are an endangered subspecies comprising three distinct pods, with just 76 individuals recorded by the Centre for Whale Research as of December 2025. The primary threats to the population are a decline in Chinook salmon, their principal food source, combined with noise pollution and vessel traffic. Bain has noted that inbreeding is also becoming a concern for a population of this size, and that the decline is likely to resume unless conditions improve materially. The combination of low population size, concentrated geographic range, and high sensitivity to acoustic disturbance makes real-time noise management one of the most direct tools available for improving the immediate living conditions of this population while longer-term challenges around prey availability and habitat quality are addressed through other policy mechanisms.

 

Read more: DRIFT Energy Secures First-Ever AiP for Energy-Harvesting Vessel Producing Green Hydrogen at Sea

 

Data Accumulation and Research Value

 

Beyond its immediate operational functions, OrcaHello is building a verified and annotated dataset of confirmed southern resident orca calls that will support future research on seasonal presence patterns, activity periods, and long-term population behaviour. The accumulation of timestamped, expert-verified acoustic records provides a scientific foundation for understanding how the orca population uses different parts of its range over time, which in turn can inform the design of shipping lane modifications, fisheries management measures, and construction scheduling frameworks. The dataset also provides material for training more sophisticated AI models capable of distinguishing between call types within the southern resident repertoire, opening possibilities for more nuanced behavioural monitoring as the system evolves.

 

Implications for Marine AI and Biodiversity Protection

 

OrcaHello illustrates how marine AI can bridge the gap between passive scientific monitoring and active operational management. Most acoustic monitoring systems generate data that is analysed retrospectively for research purposes, while OrcaHello is designed from the outset as a real-time decision support tool that connects ecological data to operational responses. The model is transferable in principle to other species and other geographic areas, and the OrcaHello team has indicated plans to scale the technology to additional locations and potentially expand it to cover other species, though the current focus remains tightly on the southern resident orcas. As AI capabilities in acoustic species recognition improve and as hydrophone networks expand, tools of this type are likely to become increasingly important components of the operational infrastructure supporting marine protected area management and endangered species recovery.

 

Outlook for OrcaHello and Real-Time Marine Conservation

 

The continued development of OrcaHello reflects a broader trajectory in conservation technology toward systems that generate actionable intelligence rather than simply data. The integration of machine learning, open-source hydrophone networks, expert moderation, and direct stakeholder alerting into a single operational pipeline represents a model that could be adapted for other marine mammals, fish aggregations, or acoustic events relevant to shipping safety and environmental compliance. For the southern resident orcas specifically, the system provides one of the most immediate and measurable interventions available given the current population status, reducing the cumulative acoustic burden imposed by coastal development and vessel traffic during critical periods when the animals are present in monitored areas.