7.8 Room modes and modal density

The previous lesson gave us the eigenfrequencies of a rectangular box: discrete modes at frequencies $f_{(n_x, n_y, n_z)}$. A real room has many such modes — and how many there are per Hz of bandwidth, the modal density, controls the perceptual character of the room at any given frequency.

Counting modes

The eigenfrequencies in the rectangular cavity all live on the lattice in $(n_x, n_y, n_z)$-space, but expressed in frequency they map to a 3-D shell in $\mathbf{k}$-space of radius $k = 2\pi f / c$. The number of modes with frequency below $f$ is the number of lattice points inside a one-octant ellipsoid in this $\mathbf{k}$-space, which for large $f$ is

where $V$ is the room volume, $S$ the total wall surface area, and $L_e$ the total length of edges. The three terms are the Weyl asymptotic expansion. The dominant volume term gives $N \propto f^3$ for large $f$.

The modal density (modes per Hz) is the derivative:

For a typical living room ($V = 50$ m³), modal density at 100 Hz is about $0.5$ modes/Hz; at 1 kHz it’s about $50$ modes/Hz; at 10 kHz, about $5000$ modes/Hz. Modes get very dense at high frequencies.

Schroeder frequency — the crossover

Each mode has some damping bandwidth $\Delta f$ — the FWHM of its response curve, set by the room’s reverberation time $T_{60}$ (the time for sound to decay 60 dB). Approximately,

For $T_{60} = 0.5$ s, $\Delta f \approx 4.4$ Hz per mode.

At low frequencies, modes are widely spaced (mode separation $\gg \Delta f$) and individually identifiable as resonances. At high frequencies, modes overlap heavily within their damping bandwidth and the spectrum looks continuous. The crossover happens at the Schroeder frequency,

where $V$ is in m³ and $T_{60}$ in seconds. For a 50 m³ room with $T_{60} = 0.5$ s, $f_S \approx 200$ Hz.

Below $f_S$: discrete modes dominate. The room “rings” at certain bass frequencies. Some places in the room are “dead” (at nodes); others “boom” (at antinodes). The perception is modal.

Above $f_S$: modes overlap into a smooth spectral response. The room is approximately a diffuse field — a statistical superposition of waves arriving from all directions with random phases. The perception is reverberant.

What modal structure does to listening

Most people have experienced modal problems without naming them:

• A specific bass note that sounds too loud in one corner of a room.
• An untreated bedroom where speech is intelligible everywhere but a single pure tone sounds different at each location.
• The “boxy” sound of small rooms — a few dominant low modes colouring everything.

The traditional architectural-acoustics treatment of these problems involves:

• Bass traps — porous absorbers placed at mode antinodes (room corners, where pressure is maximum) to damp low-frequency modes.
• Diffusers — irregular surfaces that scatter mid and high-frequency reflections into many directions, breaking up specular reflections and discrete modes.
• Geometry choice — avoiding integer-ratio dimensions (which line up many modes at the same frequencies, creating strong “ridge” resonances).

What we lose with rectangular geometry

The rectangular-cavity eigenfrequencies are an idealisation. Real rooms have:

• Furniture and bodies (additional scattering surfaces).
• Walls of finite stiffness (some absorption, some transmission to neighbouring rooms).
• Non-parallel walls (eliminating the cleanest standing-wave modes).
• Sound-absorbing surfaces with frequency-dependent impedance.

Treatment by exact eigenmodes is mostly a theoretical exercise. Practical room acoustics uses statistical descriptions (averaged mode density, decay time, reverberation) above $f_S$, supplemented by careful low-frequency mode tuning. The next lesson treats the statistical regime.