Open-Source and Low-Cost Embedded System for Frequency Domain Acoustic Monitoring

Acoustic signal analysis plays a fundamental role in predictive maintenance for industrial environments, as it enables non-invasive and continuous machine monitoring. This paper proposes a low-cost, low-power embedded system for real-time acquisition and spectral analysis of acoustic signals. The prototype uses the BitDogLab platform, an open-source educational board developed to teach electronics, programming, and embedded systems. Acoustic signals are captured via an onboard analog microphone, processed using a Hamming window and Fast Fourier Transform (FFT), and display the resulting frequency spectrum on an integrated OLED screen. Designed for flexibility and accessibility, the system enables users to perform signal processing directly on the embedded device without the need for external computing resources. As a proof of concept, the tool was used to detect spectral anomalies in the acoustic signals of a rotary machine with a planetary gear, demonstrating its practical applicability in predictive maintenance scenarios. The results highlight the system’s potential as an affordable and versatile platform for signal processing education, experimentation, and real-world applications such as anomaly detection.