### Joint news release from Julius-Maximilians-Universität Würzburg (JMU), University of Technology Chemnitz, and the Bavarian Forest National Park Authority ###
Previous apps for identifying animal calls regularly reached their limits in practice: without a stable internet connection, real-time identification in the forest or high mountains was impossible.
The new BirdNET Live app, a further development of the BirdNET app, solves this problem with a technological breakthrough: the underlying AI models have been optimised, streamlined and accelerated to such an extent that they can now run directly on the smartphone.
Around 11,000 bird species are currently known worldwide; the app can currently recognise 8,927 of them. It is also capable of identifying the calls of 268 mammals, 254 insects and 340 amphibians, in real time and completely offline. In total, BirdNET Live can identify almost 10,000 animal species by their sounds – more than any other app, as lead developer Dr Stefan Kahl from the Professorship of Media Informatics at the University of Technology Chemnitz explains: the next-best offline app can only identify around 2,300 species.
As an open-source project, BirdNET Live offers maximum usability. The app’s user interface is currently available in seven languages; the names of the identified animal species are translated into 25 languages.
A tool for experts and the general public
Whilst conventional identification apps are primarily intended for occasional use in one’s own garden, BirdNET Live is aimed first and foremost at nature conservation professionals, researchers, students and ambitious citizen scientists. Amateur ornithologists can, of course, also use it.
The app features specialised modes for structured surveys, as are standard in scientific fieldwork. It automatically records animal calls with precise location and time data and stores them locally on the smartphone.
The recordings can subsequently be flexibly reviewed, exported or shared with other experts for validation. At the same time, the app serves as a learning tool: users can listen back to their own acoustic recordings directly and thus continuously improve their species identification skills.
Deutsche Bundesstiftung Umwelt funded the development team
BirdNET Live is the result of close interdisciplinary collaboration as part of the RangerSound project, which was funded by the Deutsche Bundesstiftung Umwelt (DBU).
To identify the shortcomings of existing bioacoustic systems in everyday practical use, conservation biologists from the University of Würzburg, the rangers at the Bavarian Forest National Park, hardware experts from the company Oekofor (Freiburg / Breisgau) and AI specialists from University of Technology Chemnitz worked hand in hand.
The result is a practical and robust app designed specifically for field conditions. When used with an external microphone, for example, the smartphone can be carried safely and waterproofed in a rucksack whilst the app continuously records data in the background.
BirdNET Live is an enormous technological progress
“At first glance, running the AI locally on a smartphone seems like a small, logical step forward. However, behind this lies enormous technological progress in optimising our AI models,” says Dr Stefan Kahl from University of Technology Chemnitz, an AI expert and lead developer of BirdNET .
The app demonstrates just how important interdisciplinary collaboration is: “It was only when practitioners, conservation biologists and we, as AI developers, sat down together that we were able to identify the real shortcomings in the field and create a tailor-made, practical solution. This is a model that can be applied to many areas where AI is finding its way into our everyday lives.”
Professor Jörg Müller: “An app for students too”
“We will now be using BirdNET Live as standard in our university teaching. This will give our students entirely new opportunities to collect standardised data for their dissertations or during field courses,” says Professor Jörg Müller, Chair of Conservation Biology and Forest Ecology at the University of Würzburg.
The app’s potential extends far beyond local forests: “We expect the app to serve us exceptionally well in our research work in the extremely species-rich lowland rainforest in Ecuador, where there is no mobile phone coverage whatsoever.”
Dr Simon Verdon: “Ideal for rangers”
“Local analysis carried out directly on site is exactly what we’ve been hoping for in the day-to-day work of ranger teams in protected areas,” says Dr Simon Verdon from the University of Würzburg, a member of the RangerSound project: “They can simply leave the app running during their patrols and immediately get a clear picture of the local species composition without having to rely on mobile network coverage.”
Further facts about BirdNET
• De facto standard : BirdNET is an AI system for the automatic recognition of bird calls and the world’s leading technology for AI-supported bioacoustic monitoring.
• Successful citizen science project: With more than two million active users worldwide and over six million app downloads, more than 250 million verified nature observations have already been collected.
• Important for science : The data collected from citizen science projects feeds directly into global research projects. For example, BirdNET data collected via the BirdWeather network provided the basis for a study published in the renowned journal Nature on the impact of light pollution on birds’ singing behaviour.
• Partners : BirdNET is a joint project between University of Technology Chemnitz and the Cornell K. Lisa Yang Center for Conservation Bioacoustics (USA).
The ‘Schlaumeise’ portal helps you get started
For aspiring citizen scientists, schools, teachers and learners, the schlaumeise.org portal offers supporting materials, tutorials and project ideas covering the topics of bioacoustics, species conservation and artificial intelligence.
Further information on the app and how to download it:
• Official project website: birdnet.tu-chemnitz.de/live-app
• Portal for learners and teachers: www.schlaumeise.org