Acoustic Metamaterials for Sound Attenuation

A soundproofing panel designed and prototyped for APPL101 at UNC-Chapel Hill. The concept draws from moth wing scales, which absorb bat echolocation through an array of microscopic resonating structures. We translated this into a thin, modular panel made up of Helmholtz resonators, which consist of sealed chambers with narrow necks tuned to a target frequency. They cancel incoming sound through destructive interference and damp residual energy internally. I proposed the original concept and independently built the data analysis pipeline.

Stack

• MATLAB for all signal processing and STL computation
• Fusion 360 for prototype CAD
• 3D printing (FDM) for fabrication

Process

We built two low-fidelity prototypes: a square-chamber design and a hexagonal-chamber design, to investigate the effects of cavity geometry on tranmission impedence. Both were 3D printed as open-backed arrays of prism chambers, then sealed with a wood backing panel to complete the resonator cavity. Hole diameters were redrilled with a known-radius bit to control for printer tolerances.

To measure sound transmission loss (STL), we placed each panel at the end of a dampened cardboard tube and recorded a 20 Hz to 2 kHz frequency sweep with a microphone on the exterior face, with control recordings taken through the unobstructed tube. Each panel configuration and orientation (holes facing toward or away from the sound source) was tested twice against an average of three control recordings.

I built the analysis pipeline in MATLAB around STFT: using the Signal Processing Toolbox, a Blackman-Harris window extracts the dominant frequency at each time step, and a moving median filter locks the signal to the test tone rather than ambient noise. Averaging across repeated measurements yields dB amplitude per frequency, and subtracting control from test gives STL across the full range. I used Savitzky-Golay filtering for smoothing, with the window size chosen at the curvature knee of an absolute-difference plot via the Kneedle algorithm rather than by hand.

Outcomes

Both designs showed STL peaks near their predicted resonance frequencies, with a measured reduction of 17 dB for the hexagonal design and 15 dB for the square (compared to the control setup). These numbers could converge under our experimental uncertainty, but the small time allocation limited our ability to investigate more closely. Experimental limitations from the cardboard tube introduced standing wave artifacts and a noisy low-frequency floor, limiting our conclusions to mostly qualitative ones. However, the core result was clear, and we were able to prove that the Helmholtz resonator array concept worked as intended in physical reality.