birdnet-go

BirdNET-Go is an AI solution for continuous avian monitoring and identification. It provides 24/7 real-time bird song analysis using soundcard capture, with analysis output available to log files, SQLite, or MySQL databases.

Key Features:

• Utilizes the BirdNET AI model trained on over 6,500 bird species.
• Processes audio locally—no internet connection required.
• Includes an easy-to-use web user interface for data visualization.
• Supports over 40 languages for species names.
• Offers advanced features like Deep Detection for improved accuracy and Live Audio Streaming.
• Integrates with the BirdWeather.com API.
• Provides real-time log file output suitable for overlays in OBS (e.g., for bird feeder streams).
• Embeds the BirdNET TensorFlow Lite model directly in the compiled binary, minimizing runtime dependencies.
• Includes an endpoint for Prometheus data scraping.
• Runs on Windows, Linux, and macOS.
• Designed for low resource usage and compatibility with devices like the Raspberry Pi 3 and equivalent 64-bit single-board computers.

Target Audience: Bird enthusiasts, researchers, and conservationists interested in automated, real-time bird song identification and monitoring, especially in environments where internet access is limited or privacy is a concern.

For more details, refer to the Wiki .