10 Audio Measurement Mistakes That Ruin Your Data

Equalizing a sound system based on a measurement at one position is the most damaging common mistake. Room acoustics create a unique frequency response at every point in space. A 10 dB peak at 250 Hz at the mix position might be a 10 dB null at a seat 2 meters away. Boosting 250 Hz to fill the null makes the peak position 20 dB louder than the null position at that frequency.

Instead, measure at 4-6 representative positions across the listening area. Average these measurements (SonaVyx trace store makes this easy) and equalize based on the average. This ensures that your EQ improves the response at most positions rather than optimizing one at the expense of others.

Making EQ adjustments at frequencies where coherence is low is like navigating by a broken compass. Low coherence means the measurement is unreliable at that frequency, contaminated by noise, reflections, or measurement artifacts. Any EQ change at a low-coherence frequency will not produce the expected result.

Only make EQ adjustments at frequencies with coherence above 0.8. If coherence is low across a wide range, increase the measurement level, reduce ambient noise, or increase averaging time. If coherence is consistently low at specific frequencies, those frequencies are dominated by room reflections and cannot be effectively equalized.

Room modes create deep nulls in the frequency response at certain listener positions. These nulls occur because sound waves reflecting off parallel surfaces cancel each other. Adding EQ boost at a null frequency simply puts more energy into the room, which is still cancelled by the same reflection pattern. The null remains, but the overall level increases, wasting amplifier power and potentially damaging speakers.

Room modes can only be addressed physically: with bass trapping, speaker placement, or listening position changes. EQ can gently reduce room mode peaks (where energy accumulates) but cannot fill mode nulls.

Measuring at a level significantly different from the performance level produces misleading results. Loudspeaker response changes with level due to driver compression, port turbulence, and amplifier behavior. Room response also changes because background noise masks the measurement at low levels and nonlinear effects appear at high levels.

Measure at the expected performance SPL, typically 90-100 dB for live sound and 79-85 dB for studio monitoring. If you must measure at a lower level, increase the averaging time to improve signal-to-noise ratio and note that the measured response may differ from the performance response.

Applying too much smoothing to measurement data hides the detail you need to see. A 1-octave smoothed measurement looks reassuringly flat but conceals 10 dB peaks and dips that are clearly visible on the unsmoothed data. These hidden anomalies affect the sound quality and are the whole reason you are measuring.

Use unsmoothed data for problem identification and narrow-band corrections. Use 1/6 octave smoothing for system EQ decisions, which represents the approximate resolution of human hearing. Use 1/3 octave smoothing only for high-level spectral balance assessment. Never use more than 1/3 octave smoothing for system tuning.

Mistake 6: No windscreen outdoors. Wind noise generates massive low-frequency energy that corrupts SPL measurements. Always use a windscreen for outdoor measurements, and discard data taken in wind above 5 m/s.

Mistake 7: Covering the microphone. Holding a phone with fingers over the microphone port creates a frequency-dependent filter that adds several dB of error. Hold the phone by the edges with the microphone port unobstructed.

Mistake 8: Measuring during construction. Background noise from construction activity makes acoustic measurements unreliable. Wait for quiet conditions. Early morning or after-hours measurements are ideal for occupied buildings.

Mistake 9: Forgetting processing delay. Digital processors, DSP units, and network audio systems add latency that affects delay alignment measurements. Measure the actual acoustic delay at the listening position, not the geometric distance.

Mistake 10: Making changes without re-measuring. Every adjustment changes the system. After making EQ or delay changes, re-measure to confirm that the change had the intended effect. Without verification, you are guessing.