Room Acoustic Measurement: ISO 3382 Practical Guide

The Three Parts of ISO 3382

ISO 3382 is not a single document — it is a family of three standards, each targeting a different room type:

• ISO 3382-1:2009 — Performance spaces: concert halls, opera houses, theatres. Requires omnidirectional source, minimum 6 source-receiver combinations.
• ISO 3382-2:2008 — Ordinary rooms: classrooms, meeting rooms, offices, hospitals. Simplified procedure, minimum 2 source positions × 3 receiver positions.
• ISO 3382-3:2022 — Open-plan offices: measures spatial decay rate of speech (D2,S), A-weighted SPL of speech at 4 m (Lp,A,S,4m), and distraction distance (rD).

Most SonaVyx users work with Part 1 (venues and churches) or Part 2 (schools, offices, hospitality). The RT60 tool implements both workflows.

Generating the Impulse Response

ISO 3382 permits several excitation methods. SonaVyx supports the two most practical:

Exponential Sine Sweep (ESS / Farina Method)

A logarithmic sweep from 20 Hz to 20 kHz over 5-20 seconds. Deconvolution via the inverse filter produces the impulse response. Advantages: excellent SNR (typically 60+ dB), harmonic distortion products separated in time from the linear IR, and high immunity to background noise. SonaVyx generates the sweep in Rust WASM and performs frequency-domain deconvolution on-device.

Maximum-Length Sequence (MLS)

A pseudo-random binary sequence deconvolved via circular cross-correlation. MLS is faster than sweeps (can measure in under 1 second) but is more sensitive to time-variance (e.g., air movement, people moving) and does not separate harmonic distortion. SonaVyx implements MLS orders 2-18 with verified primitive polynomials.

Reverberation Time: T20, T30, and EDT

Schroeder Backward Integration

The Energy Decay Curve (EDC) is computed by reverse-integrating the squared impulse response: EDC(t) = ∫_t^∞ h²(τ) dτ. This is equivalent to averaging an infinite number of individual decays (Schroeder, 1965) and eliminates the statistical uncertainty of measuring individual noise decays.

Evaluation Ranges

Parameter	Evaluation Range	Min INR Required	Best For
T20	-5 dB to -25 dB	35 dB	Rooms with noise, quick surveys
T30	-5 dB to -35 dB	45 dB	Precision measurement, ISO 3382-1
EDT	0 dB to -10 dB	35 dB	Subjective perception, listener experience

T20 and T30 values are extrapolated to a 60 dB decay by multiplying the regression slope by 3 or 2 respectively. The linear regression r² value should exceed 0.995 for reliable results — the SonaVyx RT60 tool displays r² for each octave band.

When EDT Matters More Than T30

EDT is more perceptually relevant because human hearing is most sensitive to the first 10 dB of decay. In a room with early reflections (high initial energy) followed by rapid late decay, T30 might be 1.2 s while EDT is 0.7 s — the room sounds drier than T30 suggests. Conversely, a room with strong late reverberation but absorptive early conditions might have EDT > T30, sounding more reverberant than expected. See the detailed T20/T30/EDT comparison.

Room Acoustic Parameters Beyond RT60

Clarity (C50 and C80)

Clarity ratios compare early energy to late energy:

• C50 = 10 log(∫₀^50ms h²(t)dt / ∫_50ms^∞ h²(t)dt) — used for speech. Target: > 0 dB
• C80 = 10 log(∫₀^80ms h²(t)dt / ∫_80ms^∞ h²(t)dt) — used for music. Target: -2 to +4 dB for concert halls

High C50 means good speech clarity. Negative C80 means a reverberant hall suitable for organ music; positive C80 suits amplified pop/rock.

Definition (D50)

D50 = ∫₀^50ms h²(t)dt / ∫₀^∞ h²(t)dt — the fraction of energy arriving within the first 50 ms. Values above 0.50 indicate good speech intelligibility. D50 correlates strongly with STI.

Centre Time (Ts)

Ts = ∫₀^∞ t·h²(t)dt / ∫₀^∞ h²(t)dt — the time at which half the energy has arrived. Lower Ts indicates better clarity. Typical values: 40-100 ms for speech rooms, 80-200 ms for concert halls.

Measurement Positions

ISO 3382-1 requires:

• Omnidirectional sound source (dodecahedron speaker preferred)
• At least 2 source positions
• At least 3 receiver positions per source
• Minimum 6 source-receiver combinations total
• Source height: 1.5 m (approximate mouth height)
• Receiver height: 1.2 m (seated ear height)
• Minimum distance from walls: 1 m (receiver), 2 m (source)

ISO 3382-2 relaxes this to 1 source position with 3 receiver positions for ordinary rooms. The room analysis workflow in SonaVyx guides you through the position requirements.

Octave Band Analysis

All parameters should be calculated per octave band (125, 250, 500, 1000, 2000, 4000 Hz). The "single number" T30 is typically the average of 500 Hz and 1000 Hz bands. For detailed analysis, the full octave-band breakdown reveals frequency-dependent behavior — bass buildup in corners, high-frequency absorption from soft furnishings, and mid-frequency flutter echoes from parallel walls.

SonaVyx displays per-band RT60 as a bar chart with target range overlays for the selected room type. Bands outside the target range are highlighted in amber or red.

Common Pitfalls

1. Insufficient INR: If background noise is too high relative to the impulse response tail, the Schroeder curve bends upward and T30 is overestimated. SonaVyx displays INR per octave band and warns when quality thresholds are not met.
2. Source directivity: A conventional loudspeaker is not omnidirectional above 1 kHz. Results at high frequencies depend on source orientation. For rigorous ISO 3382-1 work, a dodecahedron speaker is required.
3. Air absorption: At 4 kHz and above, air absorption contributes significantly to measured RT60, especially in large rooms. ISO 3382-1 Annex A provides correction factors.
4. Time variance: People moving, HVAC cycling, or doors opening during a sweep measurement corrupt the deconvolved IR. The sweep method is more robust to this than MLS.

SonaVyx Implementation

SonaVyx implements the full ISO 3382 measurement chain in Rust WASM: sweep generation (Farina method), MLS generation (orders 2-18), frequency-domain deconvolution, Schroeder backward integration, T20/T30/EDT extraction with r² validation, and C50/C80/D50/Ts computation. The impulse response tool provides the raw IR waveform, and the RT60 tool presents the derived parameters with octave-band breakdowns and room-type target comparisons. All 20 impulse-response-related DSP tests validate against known reference values.