Treffer: Noise Spectrum Analyser

The objective of this thesis was to develop a compact Noise Spectrum Analyzer that captures, visualizes and logs environmental sound in real time. Built around an ESP32S3 and a digital MEMS microphone, the device performs onboard FFT, mapping eight frequency bands to an 8 × 8 RGB LED matrix whose column height and colour reflect amplitude. Once per minute the peak values and timestamps are published as lightweight JSON to an MQTT broker, enabling longterm, network level analysis. The firmware adopts a modular architecture, separate MicroPython modules handle audio acquisition, DSP, display, MQTT communication and user input, easing maintenance and reuse. Core algorithms were first validated on a desktop Python prototype, then ported to MicroPython and compiled with a custom ESPIDF v5.2 built in Docker to integrate nonstandard libraries. The resulting system delivers responsive visual feedback while efficiently streaming data, illustrating how low cost embedded platforms can add both aesthetic and functional acoustic awareness to smart homes, public space utilization monitoring and security applications.