Perch 2.0

Perch 2.0 is an AI model designed to help conservationists analyze bioacoustic data, specifically for those working in the field of wildlife conservation and species preservation. The tool is an update to the previous Perch model, with improved capabilities such as better state-of-the-art bird species predictions and the ability to adapt to new environments, including underwater environments like coral reefs. Perch 2.0 has been trained on a wider range of animals, including mammals, amphibians, and anthropogenic noise, making it a versatile tool for various conservation efforts.

The Perch 2.0 model works by predicting which species are present in a recording and providing tools for scientists to quickly build new classifiers starting from a single example. It also allows for vector search, which surfaces the most similar sounds in a dataset, enabling local experts to train a classifier. This process, known as agile modeling, enables the creation of high-quality classifiers in under an hour. The model can disentangle complex acoustic scenes over thousands or even millions of hours of audio data, making it an invaluable resource for conservationists.

Conservation biologists, researchers, and wildlife preservationists are among those who can derive the most value from Perch 2.0. The tool's ability to analyze vast amounts of audio data quickly and efficiently can help these professionals monitor and protect endangered species more effectively. For instance, biologists from the LOHE Bioacoustics Lab at the University of Hawaiʻi have used Perch to monitor and protect populations of honeycreepers, achieving results nearly 50 times faster than their usual methods. This accelerated process enables them to monitor more species over greater areas, ultimately contributing to the preservation of biodiversity.