Site Logo

Hello, you are using an old browser that's unsafe and no longer supported. Please consider updating your browser to a newer version, or downloading a modern browser.

Skip to main content
Environmental Solutions
January 28, 2021

Over the last fifty years scientists have developed numerous tools to monitor the health of the environment. Satellites can monitor remote areas of the planet, providing real time information on wildfires, terrestrial forest destruction, and more. But satellites are much less effective in monitoring local fauna.

Inexpensive networked field cameras can compensate for this shortcoming by monitoring animal life on the ground. But cameras can only see what is right in front of them and they are less effective in detecting small animals or those high in trees. As a result, animal populations can suffer for long periods of time unnoticed

Scientists have long recognized that recordings of animal vocalizations can monitor animal life in places that cameras cannot. As the size and cost of recorders has decreased, and the cost of data storage and processing has declined, remote audio monitoring has become an increasingly popular way of tracking certain animal populations. But for remote audio monitoring to work, processors must be able to decode animal sounds accurately.

In 2018, AVDF awarded at $250,000 grant to the Cornell Lab of Ornithology to create and refine remote acoustic monitoring tools to track ecosystem health. The project aimed, first, to develop reliable ways of detecting animal vocalizations in environments with varying background noise and, second, to train computers to identify the recorded species.

Project scientists have made significant progress on both fronts including,

–The development of a program that can correctly identify the 1,000 most common North American and European bird species as well as 5,000 additional species globally.

–The development of software that is being utilized widely by other projects, including the largest acoustic monitoring project in the U.S. utilizing 1,500 recording units now placed across the Sierra Nevada mountains.

–The launch of a spin-off project known as “KatydID” that can correctly identify 50 katydid species in Central American rainforests and is capable of being expanded as more species are recorded. This project was made possible by co-funding investment by National Geographic and Microsoft to expand the scope of the AVDF funded work.

In addition, the Lab has developed an app, BirdNET, available in 14 languages, that allows ordinary “citizen scientists” to utilize these tools. 1.1M Android app users have submitted more than 20M observations to date. The iOS version of the app was released in early December 2020 and had more than 3,000 downloads in the first month.

To learn more about the project, and to download the app, click here.


Back to all Stories