Identificirajte ptice offline pomoću AI: Nova aplikacija snima zvukove životinja direktno na vašem pametnom telefonu

A new mobile application called BirdNET Live enables users to identify animal sounds in real time and entirely offline, marking a major leap in wildlife monitoring technology. Developed through collaborative efforts involving conservationists, AI experts, and hardware engineers, the app allows smartphones to process complex auditory data without requiring an internet connection. This capability addresses long-standing limitations of previous tools, which often required stable connectivity for accurate identification. The app’s ability to function independently makes it particularly valuable for remote areas such as dense forests, mountain ranges, and protected natural reserves. The app was developed as part of the RangerSound project, which sought to enhance the effectiveness of bioacoustic monitoring in ecological research and conservation work. Researchers and rangers from the Bavarian Forest National Park played a key role in testing and refining the software, using specially designed audio loggers to collect training data. These devices were deployed in the forest for extended periods, capturing a wide range of animal vocalizations. The recorded data was used to train the AI models embedded within the app, ensuring that it could accurately distinguish between thousands of different species. BirdNET Live is based on advanced artificial intelligence models that have been significantly optimized for performance on mobile devices. Unlike earlier versions of the BirdNET app, which relied on cloud-based processing, this latest iteration runs all computations locally on the smartphone. This optimization allows the app to identify over 8,927 bird species, alongside the calls of 268 mammals, 254 insects, and 340 amphibians. In total, the app can recognize nearly 10,000 animal species by sound, a number far exceeding that of its closest competitor, which identifies approximately 2,300 species. The app’s open-source nature ensures broad accessibility and continuous improvement by the community. The app includes a variety of specialized modes tailored to both professional and amateur users. For instance, the transect mode enables users to conduct systematic surveys by recording animal calls at specific intervals along a predefined path. Each recording is tagged with precise geographic coordinates and timestamps, making it easier to analyze patterns and track changes in local biodiversity. Additionally, the app supports structured fieldwork protocols commonly used in scientific studies, allowing researchers to gather consistent and reliable data. Users can review and export their recordings directly from the smartphone, facilitating further analysis or sharing with other experts for verification. The app also functions as an educational tool, enabling users to refine their identification skills by listening to their own recordings and comparing them against the app’s database. This dual-purpose design makes BirdNET Live suitable for a wide audience, including conservation professionals, students, and citizen scientists interested in contributing to environmental research. The development of BirdNET Live highlights the importance of interdisciplinary cooperation. Conservation biologists from the University of Würzburg, rangers from the Bavarian Forest National Park, hardware specialists from Oekofor, and AI developers from the University of Technology Chemnitz collaborated to create a robust solution tailored for field conditions. The app is designed to operate reliably even in challenging environments, such as when paired with an external microphone and stored in a waterproof backpack. Continuous background recording capabilities ensure that data collection remains uninterrupted during extended field missions. With its innovative approach to offline processing and comprehensive species recognition, BirdNET Live sets a new standard for bioacoustic monitoring. As the app continues to evolve, it promises to play a vital role in advancing wildlife research and conservation efforts globally.