Surface resonance — invisible layer

experiment · accelerometer

tap the table.
your phone knows what it's made of.

Every material has a natural resonant frequency — the note it prefers to vibrate at. Steel rings at a different pitch than wood, which sounds different to glass. Lay your phone flat, give the surface a sharp tap, and the accelerometer captures the echo — a burst of vibration that tells you something about what's underneath.

Smart home devices may be able to map your floor plan from acoustic vibrations alone. Research at MIT showed that by analysing surface-conducted sound from a phone placed on different surfaces across a room, it's possible to estimate room dimensions, identify furniture arrangements, and detect wall materials — all without a microphone. The phone's accelerometer acts as a contact microphone for the surface it's resting on.

lay phone flat on a surface, then tap the surface sharply

—

detected surface type

dominant frequency: —

—peak Hz

—impact m/s²

0taps detected

accelerometer waveform

frequency content (0–50 Hz)

< 5 Hz concrete / steel very dense, little flex, fast energy dissipation

5 – 12 Hz wood / composite moderate flex, structured resonance modes

12 – 25 Hz glass / hard plastic rigid but thin — higher frequency ring-down

25 – 50 Hz thin metal / hollow thin shell resonates at higher frequencies

▶ how does surface resonance identification work?

When you tap a surface, you create a broad-spectrum impulse — a sharp burst of energy across many frequencies simultaneously. The surface then acts as a filter: it absorbs some frequencies and sustains others based on its mass, stiffness, and damping properties.

A phone accelerometer captures this as a decaying oscillation. By computing the frequency spectrum of the first 500ms after impact, we find which frequencies dominate the ring-down — the characteristic tone of the material. Dense, stiff materials (concrete, steel) dissipate energy quickly and ring at low frequencies. Light, rigid materials (glass, thin metal) ring higher and longer.

This is a simplified approximation. Professional structural health monitoring uses piezoelectric sensors and much more sophisticated signal processing, but the principle is identical.