Impulse Response: Sweep vs MLS vs Balloon Pop
TL;DR
Three methods dominate impulse response measurement: exponential sine sweep (ESS/Farina method) offers the highest SNR (60+ dB typical) and separates harmonic distortion from the linear IR, making it the gold standard per AES-2id. MLS (Maximum-Length Sequence) is faster (under 1 second for a complete measurement) and excellent for automation, but is more sensitive to time-variant conditions and does not reject distortion. Impulsive sources (balloon pops, starter pistols, hand claps) require no electronic excitation but have poor SNR (typically 30-40 dB), unrepeatable spectra, and cannot be deconvolved for distortion separation. For most measurement work, the log sweep is the recommended method. SonaVyx supports both sweep and MLS with on-device deconvolution in Rust WASM.
Why Method Choice Matters
The impulse response (IR) is the fundamental measurement from which nearly all room acoustic parameters are derived: RT60 (T20, T30, EDT), clarity (C50, C80), definition (D50), center time (Ts), and STI. The quality of the IR determines the accuracy of every derived parameter. A poor IR with 30 dB INR cannot support T30 measurement (which requires 45 dB). A time-variant measurement produces a distorted IR with artifacts that corrupt the Schroeder integration.
Exponential Sine Sweep (ESS / Farina Method)
The log sweep is a sinusoidal signal whose frequency increases exponentially from fstart to fstop over duration T. In SonaVyx, the default is 20 Hz to 20 kHz over 5 seconds at 48 kHz.
Deconvolution
The inverse filter is the time-reverse of the sweep with amplitude compensation for the logarithmic frequency distribution. Convolving the recorded signal with this inverse filter produces the impulse response. Because the sweep is logarithmic, harmonic distortion products appear at negative times (before the main IR peak) and can be windowed out — a unique advantage.
Advantages
- High SNR: 60-80 dB typical for a 5-second sweep at moderate playback levels. Longer sweeps increase SNR by 3 dB per doubling of duration.
- Harmonic rejection: Distortion products separated in time allow measurement of the purely linear IR.
- Robust to background noise: The sweep concentrates energy at one frequency at a time, improving SNR in noisy environments.
- Robust to mild time variance: A person walking during a 5-second sweep affects only a small portion of the frequency range.
Disadvantages
- Duration: 5-20 seconds per measurement plus the room's decay time. A complete measurement at one position takes 10-30 seconds.
- Audibility: The sweep is clearly audible — measurement during an event is impractical.
- Single-shot: Each measurement requires a full sweep playback. Cannot be averaged in real-time.
Maximum-Length Sequence (MLS)
An MLS is a pseudo-random binary sequence (±1) with the property that its circular auto-correlation is a near-perfect impulse. The IR is extracted via circular cross-correlation between the played MLS and the recorded signal.
Advantages
- Speed: An MLS of order 16 (65,535 samples = 1.37 s at 48 kHz) completes in under 2 seconds. Multiple sequences can be averaged for higher SNR.
- Broadband excitation: Energy is distributed evenly across all frequencies simultaneously.
- Automation-friendly: Short duration enables rapid multi-position surveys.
Disadvantages
- Time-variance sensitivity: Any change during the sequence (movement, HVAC) corrupts the cross-correlation. This is the primary practical limitation.
- No distortion separation: Harmonic distortion products appear in the recovered IR as artifacts, potentially corrupting RT60 calculations.
- Lower SNR per unit time: A 1.4-second MLS has approximately 15 dB less SNR than a 5-second sweep.
Impulsive Sources
Balloon pops, starter pistols, and hand claps generate broadband transients that approximate an impulse. The captured waveform IS the impulse response (no deconvolution needed).
Advantages
- No electronic source required: Useful when no PA system is available.
- Simple: Pop a balloon, record the result.
- Fast: Instantaneous excitation.
Disadvantages
- Poor SNR: 30-40 dB typical — insufficient for T30 measurement in most rooms.
- Unrepeatable spectrum: Each balloon pop has a different spectral shape, making averaging problematic.
- Limited low-frequency energy: Balloon pops have poor energy below 100 Hz. Starter pistols are better but raise safety concerns.
- Crest factor: The impulsive nature can overdrive microphone preamps if gain is set for normal speech levels.
Comparison Table
| Parameter | Log Sweep | MLS | Balloon Pop |
|---|---|---|---|
| SNR (typical) | 60-80 dB | 45-60 dB | 30-40 dB |
| Duration | 5-20 s | 1-2 s | Instant |
| Distortion rejection | Yes | No | No |
| Time-variance robust | Good | Poor | Excellent |
| Deconvolution needed | Yes | Yes (cross-corr) | No |
| T30 capable (45 dB INR) | Yes | With averaging | Rarely |
| Equipment needed | Speaker + mic | Speaker + mic | Balloon + mic |
SonaVyx Implementation
The impulse response tool supports both sweep (default) and MLS excitation. The sweep generator produces a configurable-duration log sweep (2-20 seconds, 20 Hz-20 kHz). Deconvolution uses frequency-domain convolution with the pre-computed inverse filter. MLS generation uses verified primitive polynomials for orders 2-18. Both methods are implemented in Rust WASM with 20 dedicated unit tests.
For the best results, use the sweep method with a duration of at least 5 seconds. The room analysis workflow defaults to a 10-second sweep for maximum SNR. If time is critical (multi-position survey), switch to MLS with 4× averaging for a practical balance of speed and accuracy.
After capturing the IR, derived parameters are computed automatically: RT60 (T20, T30, EDT) per octave band, C50, C80, D50, Ts, and INR quality indicators. The AI diagnostic interprets these parameters in the context of your venue type.
Try It Now
Open this measurement tool in your browser — free, no download required.
Frequently Asked Questions
Last updated: March 19, 2026