Talk2NatureMarine mammals / Project announcement
DolphinGemma and shared signals
Source year 2025 · Note reviewed 2026-10-04
DolphinGemma models patterns in dolphin sounds. The related CHAT work uses learned shared signals in human-dolphin interactions.
What we reviewed · Initial source note
Official announcement, including its field-program location paragraph, reviewed; not an independent replication or complete dataset audit.
Not yet independently reviewed by a scientific specialist.
What this does — and doesn’t — establish
Learning a shared interface and translating spontaneous natural communication are distinct scientific tasks.
What we can build on
Choose which question a pilot is trying to answer before selecting the interface or model.
See the related tools and reading paths →
The next reading task
Keep natural-call interpretation and taught-symbol experiments in separate protocols.
Go to the sources
- DolphinGemma ↗
Google / Herzing & Starner. DolphinGemma announcement (14 April 2025). Natural sound modeling and the separate CHAT project for learned shared signals.
Original editorial note by Talk2Nature contributors. AI-assisted, source-grounded synthesis; not a peer review or an independent replication. Linked works retain their own rights.