An IoT system that listens for dog barks using an ESP32 microcontroller and surfaces noise levels as observability metrics in Dynatrace.

The ESP32 samples audio from an analogue microphone and computes RMS amplitude over 100-sample windows. Each measurement is timestamped via NTP and published to an MQTT broker. A Go backend subscribes to the broker, parses the payload, and exports a bark_power gauge metric through the OpenTelemetry SDK to an OTel Collector — from where it is forwarded to Dynatrace for dashboarding and alerting.

The backend and Mosquitto broker run as containers on a Raspberry Pi cluster at home, making it a fully self-hosted, production-grade setup on low-power ARM hardware.

Check it out here.