DolphinGemma: How Google AI is helping decode dolphin communication

Google has introduced DolphinGemma, an AI model developed to analyze and interpret dolphin vocalizations. Collaborating with the Wild Dolphin Project (WDP) and Georgia Tech, the initiative aims to understand the structure and potential meanings of dolphin sounds. DolphinGemma utilizes Google's SoundStream tokenizer and is optimized to run on Pixel smartphones, facilitating real-time analysis in the field. The model processes sequences of natural dolphin sounds to identify patterns and predict subsequent sounds, akin to how language models predict the next word in a sentence. Additionally, the CHAT (Cetacean Hearing Augmentation Telemetry) system, integrated with Pixel devices, enables two-way interaction by associating synthetic whistles with specific objects, allowing dolphins to "request" items. Google plans to release DolphinGemma as an open model, potentially aiding researchers studying various cetacean species.

Source and image: https://blog.google/technology/ai/dolphingemma/