AES-2id: SNR Requirements and Quality Verification
TL;DR
AES-2id emphasizes SNR as the primary quality metric for impulse response measurements. The effective SNR (or INR per ISO 3382-1) must exceed the evaluation range plus a safety margin: at least 45 dB for T30, 35 dB for T20, 25 dB for EDT. AES-2id recommends verifying SNR by comparing the impulse response tail level to a noise-only capture. The document also introduces the concept of "effective SNR" for swept sine measurements: the processing gain from deconvolution adds 10·log10(BT) dB where B is bandwidth and T is sweep duration.
SNR as Quality Metric
The usable dynamic range of a measured impulse response is determined by the signal-to-noise ratio. AES-2id aligns with ISO 3382-1 INR thresholds but provides additional guidance on how to achieve and verify adequate SNR in practice.
Effective SNR for Swept Sine
The swept sine method provides processing gain through correlation. The effective SNR after deconvolution exceeds the instantaneous SNR during playback by:
Processing gain = 10 · log₁₀(B · T) dB
Where B is the sweep bandwidth (Hz) and T is the sweep duration (seconds). For a 20 Hz – 20 kHz sweep of 10 seconds: gain = 10·log₁₀(20000 × 10) ≈ 53 dB. This explains why swept sine measurements can achieve 60+ dB of usable dynamic range even in moderately noisy environments.
Verification Method
AES-2id recommends the following verification procedure:
- Capture a noise-only recording (no excitation signal) with the same recording chain
- Apply the same deconvolution processing to the noise-only recording
- Compare the level of the deconvolved noise to the impulse response tail
- The difference is the effective SNR per frequency band
This method captures all noise sources: ambient room noise, electronic noise in the recording chain, and quantization noise.
Per-Band Assessment
SNR should be assessed in each octave band independently, as it varies significantly across frequency:
| Band | Typical challenge |
|---|---|
| 63 Hz | High ambient noise, limited source output |
| 125 Hz | HVAC rumble, traffic |
| 250 Hz – 2 kHz | Usually best SNR |
| 4 kHz | Air absorption in large rooms |
| 8 kHz | High air absorption, low source output |
Improving SNR
When initial measurement shows insufficient SNR:
- Increase sweep duration (each doubling adds 3 dB)
- Increase source output level
- Reduce background noise (turn off HVAC, close doors)
- Average multiple sweeps — see averaging strategies
- Reduce electronic noise (use a quality audio interface, avoid long cable runs)
Clipping and Overload
While increasing level improves SNR, overloading the recording chain introduces harmonic distortion that contaminates the impulse response. AES-2id recommends setting the recording level to peak at -6 dBFS to -3 dBFS, leaving headroom for unexpected transients.
SonaVyx Quality Display
The IR tool computes and displays per-band INR quality after each measurement using color-coded indicators (green ≥45 dB, amber 35-44 dB, red <35 dB). If INR is insufficient, the tool suggests specific improvements. For common measurement errors that degrade SNR, see the troubleshooting section.
Try It Now
Open this measurement tool in your browser — free, no download required.
Last updated: March 19, 2026