AI's Next Frontier: Decoding Dolphin Communication

By Kate Taylor | Science | November 5, 2026

Denise Herzing, founder and director of the Wild Dolphin Project, is using AI to analyze audio and video footage of the marine mammals—bringing us closer to understanding their complex communication.

The Quest to Understand Dolphin Speech

For decades, scientists have wondered what dolphins are saying to one another. Now, advances in artificial intelligence are offering an unprecedented opportunity to find out. Denise Herzing, founder and director of the Wild Dolphin Project, is at the forefront of this effort, employing AI to sift through vast amounts of audio and video recordings of dolphins in the wild.

Herzing's work represents a growing intersection of marine biology and machine learning—a field that has accelerated rapidly as computational tools have become more powerful and accessible. By training AI models on dolphin vocalizations and associated behaviors, researchers hope to uncover patterns that have eluded human observers.

Why AI Is Essential

Dolphins produce a wide range of sounds—whistles, clicks, and burst-pulse calls—that vary in frequency, duration, and context. Analyzing these signals manually is painstaking and time-consuming. AI can process thousands of hours of recordings quickly, identifying recurring patterns and correlating them with specific behaviors captured on video.

As of 2026, large language models and multimodal AI systems have become adept at handling complex, noisy datasets. These same technologies are now being adapted to decode non-human communication, opening new avenues for interspecies understanding.

Looking Ahead

The Wild Dolphin Project's work is still ongoing, but the implications are profound. If AI can help us decipher dolphin communication, it could transform our understanding of animal intelligence and lead to more effective conservation efforts. For now, Herzing and her team continue to collect data and refine their models—one whistle at a time.

This article was originally published on November 5, 2026.

via Wired AI