Google is diving deep into the ocean to better understand dolphin communication—and it’s using the Pixel 9 and AI to do it.

Dolphin Conversations Go High-Tech

The Wild Dolphin Project (WDP), a nonprofit studying Atlantic spotted dolphins, has been tracking dolphin behavior for decades. Now, with help from Google’s AI model called Dolphin Gemma, they’re closer than ever to “talking” with them.

Dolphin Gemma is based on Google’s open-source Gemma series and is compact enough to run directly on mobile devices. Trained on WDP’s extensive library of underwater audio, the model can produce dolphin-like vocalizations and soon, may even generate real-time replies. That’s right—Google aims to create two-way interactions between humans and dolphins.

Previously, WDP researchers relied on Pixel 6 smartphones to collect and analyze dolphin sounds. The upcoming switch to the Pixel 9 marks a major leap. With more power and improved efficiency, the Pixel 9 will allow researchers to run both AI and sound-matching algorithms simultaneously—no bulky gear required. This mobile-first approach streamlines field research, especially in the open ocean where size, power consumption, and portability matter.

What Dolphin AI Is Actually Listening For

Using underwater recording, WDP researchers can match specific sounds with behaviors they observe in real time. Some of these vocal patterns include:

By decoding these vocal cues, researchers hope to uncover deeper structures in dolphin chatter—possibly even elements that resemble language.

The Bigger Picture

One exciting application is through CHAT (Cetacean Hearing Augmentation Telemetry), a system Dolphin Gemma could boost by predicting vocal patterns earlier in interactions. This means researchers might soon respond to dolphin communication more fluidly, creating smoother back-and-forth exchanges.

With AI and the Pixel 9 working together, what once seemed like science fiction—human-dolphin communication—could soon be a real-world breakthrough.