Skip to content
All projects
EmbeddedML

SonicSense

Dual-modal acoustic anomaly detection running on-device on ESP32-S3.

Inference
on-device TinyML
Compute
ESP32-S3

Overview

An on-device acoustic anomaly detection system using dual-modal sensing and TinyML inference, running entirely on an ESP32-S3 with no cloud dependency.

Approach

Combines two acoustic input modalities to increase detection robustness. A quantized anomaly-detection model runs on-device, enabling real-time classification within the ESP32-S3's memory and compute constraints.