PulseAugur
LIVE 13:09:30
research · [1 source] ·
0
research

Google's bird-trained AI excels at identifying whale vocalizations

Google DeepMind has released Perch 2.0, a bioacoustics foundation model primarily trained on terrestrial animal vocalizations, which demonstrates surprising effectiveness in underwater acoustic analysis. Despite not being trained on marine sounds, Perch 2.0 achieved strong performance in classifying whale vocalizations through transfer learning. This advancement, detailed in a NeurIPS workshop paper, enables more efficient insights into marine ecosystems and includes a tutorial for custom classifier development using Google Cloud. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

RANK_REASON The release of a specialized AI model and accompanying research paper detailing its novel application falls under the 'research' bucket.

Read on Google AI / Research →

Google's bird-trained AI excels at identifying whale vocalizations

COVERAGE [1]

  1. Google AI / Research TIER_1 ·

    How AI trained on birds is surfacing underwater mysteries

    Climate & Sustainability