PERCH 2.0 — helps to figure out which bird sings or who croaks

perch 2.0 arose at the junction of artificial intelligence and environmental protection. It offers tools to solve some acute environmental problems. These are in the field of biocoustics.

Google Deepmind Perch clearly demonstrates the deep influence of specialized AI on the development of science, in particular in biocoustics, writes Xrust. Nature defenders faced a difficult task: to comprehend vast audio -dedicated, collected in wild ecosystems. These notes, saturated with the sounds of birds, frogs, insects, whales and fish, give invaluable information about the presence of animals and the state of ecosystems.

However, manual analysis of millions of hours of audio recordings is a monumental task. It is here that Perch, an artificial intelligence model intended for the analysis of biocoustic data, comes to the rescue. The updated PERCH 2.0 model is a significant step forward, offering more modern ready-made predictions of bird species than its predecessor.

It is important to note that it can more effectively adapt to new conditions, including difficult underwater conditions, such as coral reefs. Its training set of data is almost twice as much as the previous version, and includes a wider range of animal vocalization, including mammals and amphibians, as well as anthropogenic noise from public sources, such as Xeno-Canto and Inaturalist.

commitment to the principles of open science is obvious: Perch 2.0 has an open source code and is available on Kaggle, which contributes to a wide implementation in the scientific community. Since the launch in 2023, the first version of PERCH was downloaded more than 250,000 times, integrating open source solutions into tools for practicing biologists, such as Birdnet Analyzer Cornell.

Xrust PERCH 2.0 — helps to figure out which bird sings or who croaks