[vlovich123]

They’ve been working on decoding dolphin sounds for a long time - Thad was telling me about this project in 2015 and it had been ongoing for a while. One challenge is doing this realtime is extremely difficult because of the frequency the dolphin speech occurs in. And they want to do this realtime which adds to the difficulty level. The other challenge on the AI side is that traditional AI is done using supervised learning whereas dolphin speech would require unsupervised learning. It would be interesting to learn more about how Gemma is helping here.