Google releases enhanced version of AI 'Perch' that can estimate 'the type and number of living things in the surrounding area' from environmental sounds



Google's AI, Perch , is a pre-trained model for bioacoustics that analyzes environmental sounds to answer a variety of questions, such as how many animals are in an area and how many babies have been born. Google announced a new version of Perch on August 7, 2025, which is more powerful than ever, including the ability to recognize more animal sounds and interpret the sounds of underwater environments.

New Perch AI model helps protect endangered species

https://blog.google/technology/google-deepmind/perch-ai-model/

How AI is helping advance the science of bioacoustics to save endangered species - Google DeepMind
https://deepmind.google/discover/blog/how-ai-is-helping-advance-the-science-of-bioacoustics-to-save-endangered-species/

[2508.04665] Perch 2.0: The Bittern Lesson for Bioacoustics
https://arxiv.org/abs/2508.04665

Perch is an AI model specializing in bioacoustics released by Google DeepMind in 2023. It initially had analytical capabilities specialized in bird calls and was able to distinguish the calls of more than 10,000 different bird species from environmental sounds.

Perch 2.0, announced on August 7, 2025, expands the environmental sounds that Peach can analyze to include not only birds but also mammals and amphibians, and can also analyze ocean and artificial sounds.

The uses and mechanisms of Peach are explained in the following video.



Animals communicate their locations by making cries and other sounds. Peach is an AI that can extract these sounds from environmental sounds to determine what animals are there, where they are located, and how many of them there are.



By recording environmental sounds using equipment installed in nature and analyzing the distribution and population of animals using Peach, it is possible to analyze the impact of climate change on ecosystems and the impact of human settlement expansion on ecosystems.



Some animal cries are so similar that they are almost indistinguishable to the human ear. Peach uses AI-based acoustic analysis to make it possible to distinguish between individuals that were previously difficult to distinguish. It can also distinguish between cries made by adults and babies.



Peach has been downloaded more than 250,000 times since its release in 2023, and its search library is part of Cornell University's widely used BirdNet Analyzer. It has also helped Australian bird conservation organizations and the Australian Acoustic Observatory develop tools to classify many species, enabling the discovery of new populations that were previously difficult to find. Google DeepMind says, 'We expect the new version of Peach to further accelerate these efforts.'

Peach 2.0 is available for download and use on Kaggle , an online platform for data science and machine learning.

Google | Perk | Kaggle
https://www.kaggle.com/models/google/bird-vocalization-classifier/tensorFlow2/perch_v2

in AI,   Software, Posted by log1e_dh