BirdNET was developed by Chemnitz University of Technology and Cornell University (USA). It identifies 3000 bird species by their song using an artificial intelligence algorithm.
Recognizing bird songs is a complex process. The core of the BirdNET application is an artificial intelligence-based algorithm connected to a server at Chemnitz University of Technology. The artificial neural network determines the recorded voice of the bird upon request. With this in mind, the researchers set out to make sure that the application’s speed would not be affected by the number of users – and therefore search queries. The developers claim that they succeeded.
Following the completion of a test run with a research team in India, the identification rate of native bird species ranged from 80% to 85%. This value can hardly be increased, because when recording the calls of birds in the open air, various ambient noises are often heard, the scientists explain. But in most cases, the application will recognize the corresponding bird cry without problems.
BirdNET application at a glance
Task: automatic recognition of bird calls based on short audio recordings. The application analyzes the sound wave of the bird cry, as well as the location and date of the recording. Based on the data, the AI decides whether a particular species of bird can be heard. After the species has been identified, detailed information is provided for each bird. You can save your observations and share with friends. Each observation is recorded anonymously and evaluated for research purposes. User feedback is constantly incorporated into the design of the application and the features implemented.
The app is designed to help people get to know their environment better and, ideally, raise their environmental awareness. However, it is useful not only for recreation, but also for ornithological research, scientists conclude. It promotes awareness and research on biodiversity in nature.
