Acoustic sensor node for HomeAssistant, Meshtastic, Websocket, and APIs. Online and offline mesh capabilities. ESP32-S3 with three PMD mics, Lora module, GPS module, compass module, power module, and battery.

Audio Pipeline: energy gate and gain setting - raw audio data is classified with Google’s Yamnet model (>500 classes) and weights, various filter settings (e.g., background noise, confusion cluster, threshold and cooldown).

Learning: Custom scene fingerprints (sound fingerprint over time) and sounds can be learned.

Broadcast: Scene triggers, modes and filtered stream can be broadcasted to HomeAssistant, Meshtastic Channel. Mesh functions.