How AI advances biological science to save endangered species

Our new perch model helps conservationists analyze audio faster to protect endangered species, from Hawaii’s honey cliffers to coral reefs.

One way scientists can protect the health of their planet’s wild ecosystems is to use microphones (or underwater hydrophones) to collect huge amounts of audio densely packed with vocalizations from birds, frogs, insects, fish, and more. These recordings can tell you a lot about animals present in a particular area, along with other clues about the health of their ecosystem. However, understanding so much data remains a large-scale effort.

Today we are releasing an update for Perch. Perch is an AI model designed to help conservators analyze bioacoustic data. This new model has better cutting edge off-the-spot bird species predictions than its previous models. It can be adapted to underwater environments, especially underwater environments, such as coral reefs. They are trained on a wider range of animals, including mammals, amphibians and anthropogenic noise. This is almost twice as much data from public sources such as Xeno-Canto and Inaturalist. Unleash complex acoustic scenes across thousands or millions of hours of audio data. And it is versatile and can answer a variety of questions, from “the number of babies born” to “the number of individual animals present in a particular area.”

To help scientists protect our planet’s ecosystem, we have sourced this new version of the perch and made it available at Kaggle.

Success Story: Perch on the Field

Since its first launch in 2023, the early version of Perch has already been downloaded over 250,000 times, and its open source solution is now integrated into a tool for work biologists. For example, Perch’s Vector Search Library is part of Cornell’s widely used Birdnet Analyzer.

Additionally, Perch supports many unique Australian species Birdlife Australian and Australian Acoustic Observatory Build Classifiers. For example, our tools have enabled the discovery of a new population of Wanderer in the elusive plains.

Recent studies have found that previous versions of perches can be used to identify individual birds, track bird abundance, and potentially reduce the need for catch-and-release studies to monitor populations.

Finally, biologists at the Lohe Bioacoustics Lab at the University of Hawaii used it to monitor and protect a population of honey cliffers facing extinction from the threat of bird malaria spread by non-native mosquitoes, which is important to Hawaiian mythology. Perch helped Lohe Lab find honey cliffer sounds 50 times faster than the usual way, allowing more honey cliffers to be monitored in larger areas. We hope that the new model will further accelerate these efforts.

Unleash the planet playlist

The perch model can predict which species will be present in the recording, but that is only part of the story. It also provides an open source tool that allows scientists to quickly build new classifiers starting from a single example, allowing them to monitor new classifiers for very specific sounds like juvenile calls. Considering one example of a sound, vector search using perch surfaces is the most similar sound in the dataset. Local experts can mark search results as relevant or irrelevant to train classifiers.

Together, the combination of vector search, active learning and powerful embedded models is called agile modeling. Our recent paper, “Squawk: Agile Modeling in Bioacoustics” – shows that this method works between birds and coral reefs, allowing you to create high-quality classifiers within an hour.

Looking ahead: The future of biological acoustics

Together, our models and methods help maximize the impact of conservation efforts, leaving more time and resources for meaningful ground work. From Hawaiian forests to sea forests, the Perch Project illustrates the profound impact we can have when applying technical expertise to the world’s most pressing challenges. All the classifiers constructed and hourly data analyzed bring us closer to a world where our planet’s soundtrack is one of the rich and thriving biodiversity.