Today, on National Dolphin Day, Google, in collaboration with researchers at Georgia Tech and the field research of the Wild Dolphin Project (WDP), is announcing progress on DolphinGemma: a foundational AI model trained to learn the structure of dolphin vocalizations and generate novel dolphin-like sound sequences. This approach in the quest for interspecies communication pushes the boundaries of AI and our potential connection with the marine world.
pull down to refresh
0 sats \ 2 replies \ @JesseJames 14 Apr
Dolphins are smarter than us... :-)
reply
5 sats \ 1 reply \ @carter OP 14 Apr
I can't wait to talk to them... I wonder if we can like hire them to do jobs for fish
reply
31 sats \ 0 replies \ @teemupleb 14 Apr
When will we have the first dolphin on the Joe Rogan podcast?
reply
100 sats \ 0 replies \ @StackerJack 14 Apr
reply
0 sats \ 0 replies \ @Dkryptoenth 8h
Actually, Google AI has been involved in research projects aimed at decoding animal communication, including dolphin vocalizations, as part of broader efforts like Project CETI (Cetacean Translation Initiative). Here’s how AI, especially from organizations like Google, helps in detecting and interpreting dolphin "language":
- Data Collection
Dolphins communicate using clicks, whistles, and other sounds.
Researchers use underwater microphones (hydrophones) to record thousands of hours of dolphin vocalizations.
- Signal Processing
AI algorithms help clean and process noisy underwater recordings to isolate dolphin sounds from background noise like waves or boat engines.
- Pattern Recognition
Machine learning models, especially deep learning, are trained to recognize patterns in the dolphin vocalizations.
These models can detect specific types of clicks or whistles and correlate them with dolphin behavior, like hunting or socializing.
- Translation Efforts
Natural Language Processing (NLP), the same tech behind Google Translate, is adapted to look for possible syntax or structure in dolphin communication.
AI attempts to map sounds to meanings or contexts, essentially creating a primitive “dictionary” of dolphin sounds.
- Unsupervised Learning
Since we don’t "speak dolphin," unsupervised learning helps the AI find structure in the data without needing labeled examples.
Clustering algorithms group similar sounds together, which can hint at repeated phrases or "words."
- Real-Time Detection
Advanced AI systems can eventually be used for real-time monitoring of dolphin pods to understand how they respond to changes in their environment.
This field is still in early stages, but with the help of powerful AI tools and big data, researchers are getting closer to understanding the complexity of dolphin communication.
reply
0 sats \ 0 replies \ @Dkryptoenth 14 Apr
There ain't nothing technology can't achieve.
reply