ISO 3382-1 Impulse-to-Noise Ratio (INR) Requirements
TL;DR
The impulse-to-noise ratio (INR) is the difference in decibels between the peak of the impulse response and the noise floor. ISO 3382-1 Clause 8.3 requires a minimum INR of 35 dB for T20 extraction and 45 dB for T30 extraction. Insufficient INR causes the Schroeder integration to curve upward at the tail, biasing the regression slope and overestimating reverberation time. INR is checked per octave band — low-frequency bands often have the worst INR due to higher ambient noise and lower source output. Multiple averages improve effective INR by 3 dB per doubling.
INR Definition and Thresholds (Clause 8.3)
INR measures the usable dynamic range of the measured impulse response. It is defined as the difference between the peak level of the impulse response and the steady-state background noise level, both in the same octave band:
INR = Lpeak,IR - Lnoise (dB)
ISO 3382-1 specifies minimum INR thresholds:
| Parameter | Required INR | Reason |
|---|---|---|
| T20 | ≥ 35 dB | Regression uses -5 to -25 dB range (20 dB span + 10 dB margin) |
| T30 | ≥ 45 dB | Regression uses -5 to -35 dB range (30 dB span + 10 dB margin) |
| EDT | ≥ 25 dB | Uses 0 to -10 dB range (10 dB span + margin) |
The 10 dB margin above the regression range ensures the noise floor does not contaminate the decay curve within the evaluation region.
Consequences of Insufficient INR
When the impulse response tail merges with the noise floor, the Schroeder backward integration curves upward instead of continuing linearly downward. A linear regression fitted to this contaminated region yields a shallower slope, resulting in an overestimated reverberation time. The error can be several tenths of a second — significant for precision measurements.
Similarly, C80 and D50 are affected because the noise floor adds energy to the late portion of the integral, reducing the apparent clarity.
Improving INR
Strategies to achieve sufficient INR:
- Increase source level — use a more powerful amplifier/loudspeaker combination
- Use swept sine excitation — provides 10-20 dB more effective SNR than impulsive sources for the same peak level
- Average multiple measurements — synchronous averaging of N impulse responses improves INR by 10·log₁₀(N) dB (3 dB per doubling)
- Reduce background noise — turn off HVAC, close doors, measure during quiet periods
- Use longer sweep duration — doubling sweep length adds 3 dB to effective SNR
Per-Band Verification
INR must be verified in each octave band independently. Low frequencies (63 Hz, 125 Hz) often have the worst INR because ambient noise tends to be highest at low frequencies while source output may be limited. High frequencies (4 kHz, 8 kHz) can also be problematic in large rooms due to air absorption reducing the impulse response level.
Lundeby Truncation
The Lundeby method automatically detects where the impulse response meets the noise floor and truncates the integration at that point, compensating for the upward tail bias. SonaVyx applies Lundeby truncation by default in the RT60 tool.
SonaVyx INR Reporting
SonaVyx computes and displays INR quality for each impulse response measurement. The quality indicator uses ISO 3382-1 thresholds: green (≥45 dB, T30 quality), amber (35-44 dB, T20 only), red (<35 dB, insufficient). This helps you decide during measurement whether additional averaging or level increase is needed. See uncertainty reporting for how INR affects measurement confidence.
Try It Now
Open this measurement tool in your browser — free, no download required.
Last updated: March 19, 2026