Sound travels strangely underwater, and reading it takes high-level skill. Fortunately, a major Japanese tech company wants to teach a computer to do that reading for us.

NEC Corp just landed a contract to build an AI model that can make sense of underwater acoustic data. 

The project falls under a Japanese defense research program, and it could change how we study the ocean.

NEC

NEC is headquartered in Tokyo. Over the next two years, the company will design and train what’s called a foundation model. 

This will happen through Japan’s Defense Innovation Science and Technology Institute. That group sits under the country’s Agency for Defense Equipment. 

The research has a formal name too: Research on Underwater Acoustic Foundation Models Using Self-Supervised Learning, Part 1.

Underwater Sound

Sound behaves differently underwater than it does in air. Waves bend and bounce off different water layers, temperatures, and pressures. 

That makes patterns tough to spot, even for trained experts. Most AI models today need huge amounts of labeled data to learn well. 

That’s the problem here: labeled underwater acoustic data is scarce and hard to collect. Therefore, NEC says its model will use something called self-supervised learning. 

This lets the AI learn patterns on its own, without needing every piece of information tagged by a human first.

Cotomi Core

NEC plans to use its own AI technology, which includes its Cotomi core AI system, for this build. The company will also lean on large-scale underwater acoustics data it has gathered over time.

The goal sounds ambitious. NEC wants to create something similar to a large language model, but built specifically for ocean sound instead of human text. 

In simple terms, they’re trying to give AI the ability to “listen” to the sea and understand what it hears. 

Digital Twin

NEC hopes this technology can help build a digital twin of the oceans. That means a detailed, data-driven model that mirrors real ocean conditions and behavior.

With that kind of tool, scientists and defense officials could better track subsea phenomena. This includes anything from underwater movement to hidden geological changes.

Geological Benefits

The uses for this AI stretch far beyond defense. NEC listed several possible applications, and they cover a lot of ground.

High-precision earthquake prediction sits near the top of the list. Better acoustic data could help researchers spot warning signs earlier. 

Marine life and resource exploration could also benefit, giving scientists new ways to study creatures and materials deep underwater.

Finally, scientists could use the AI for environmental monitoring within the context of climate change. Tools like this could help track changes more accurately over time.