Our bioacoustics foundation model Perch was trained primarily on terrestrial animals - like birds. But Perch 2.0 is showing incredible performance on underwater acoustics. Here’s how it’s helping us listen to and understand marine ecosystems 🧵

The ocean is a “soundscape” of mystery. To protect underwater species, we need to identify them at scale. Recently, we launched Perch 2.0, which excels at marine validation tasks - despite having no underwater audio in training.

To extend the capabilities of Perch 2.0 for underwater sound identification, we leveraged transfer learning. The pre-trained model can already understand sound, so we teach it to learn new parameters only for the final step in the process, applying it to a new species.

Perch 2.0 was evaluated on a range of whale vocalization tasks: distinguishing different baleen whale species, and killer whale subpopulations. Compared against pre-trained models, it landed consistently in the top or second-best performing model for each dataset and sample size.

We believe Perch 2.0 is so successful at transferring bird data to underwater sounds because of its ability to: Generalize: it can expertly categorize sounds not included in its training. Classify similar sounds: it’s forced to learn detailed acoustic features. Recognise diversity: identify a variety of species that have evolved vocalizations, yet still share structural characteristics.