Acoustic Field Measurement Guide — RT60, STI, BGN With Real Equipment

68% of post-completion acoustic surveys reveal at least one room that does not meet its design target, according to a 2024 study by the Association of Noise Consultants across 340 UK commercial buildings. In 23% of those cases, the discrepancy was caused not by poor design but by incorrect measurement — the surveyor used the wrong method, insufficient measurement positions, or equipment that was not calibrated. An acoustic measurement is only as good as the methodology behind it.

This guide covers the practical workflow for measuring the three most commonly specified acoustic parameters in the field: reverberation time (RT60) per ISO 3382-2:2008, speech transmission index (STI) per IEC 60268-16:2020, and background noise level per ISO 1996-2:2017. It is written for acoustic consultants, building surveyors, facilities managers, and architects who need to conduct or commission acoustic measurements and understand whether the results are reliable.

Part 1: Equipment Selection

Sound Level Meters

The foundation of all acoustic measurement is the sound level meter (SLM). ISO 3382-2 and IEC 60268-16 require a meter conforming to IEC 61672-1:2013 Class 1 (the highest accuracy class) for compliance measurements.

Critical note: Class 2 meters (IEC 61672-1 Class 2) have wider tolerance bands (±2.3 dB vs ±1.1 dB at reference frequency) and are not acceptable for ISO 3382-2 compliance measurements. Many budget SLMs marketed as "professional" are Class 2 — always verify the classification.

Sound Sources

Omnidirectional loudspeakers: ISO 3382-2 §5.3 requires a sound source with maximum deviation from omnidirectional radiation of ±1 dB (averaged over 30° arcs in any plane) at frequencies up to 5000 Hz. Dodecahedral loudspeakers (12 drivers arranged on the faces of a dodecahedron) meet this requirement.

Power amplifiers: The loudspeaker must be driven by an amplifier capable of delivering sufficient power to achieve a signal-to-noise ratio of at least 45 dB (for T30) or 35 dB (for T20) above the background noise in each octave band. A 200–500 W amplifier is typical.

Impulsive sources: For survey-grade measurements, a balloon pop (diameter ≥ 300 mm, inflated to bursting) or starter pistol can substitute for a loudspeaker. Impulsive sources provide adequate dynamic range for T20 measurements in most rooms but have limited low-frequency energy (balloon pops are weak below 125 Hz) and are not repeatable.

Calibration

Every measurement session must begin and end with calibration using a sound calibrator conforming to IEC 60942:2017 Class 1 (e.g., Brüel & Kjær Type 4231: 94 dB at 1000 Hz, or Norsonic Nor1255: 94/114 dB at 1000 Hz). The calibration check verifies that the SLM's sensitivity has not drifted. A pre-to-post calibration difference exceeding ±0.5 dB invalidates the measurement session.

Part 2: Measuring Reverberation Time (RT60) per ISO 3382-2:2008

Method 1: Interrupted Noise

Procedure:

1. Set up the dodecahedral loudspeaker at the first source position (typical use position — lectern, presenter position, conductor position). Height: 1.5 m above floor.
2. Connect the pink noise generator and amplifier. Adjust the output level to achieve at least 45 dB above background noise in each octave band at the furthest receiver position.
3. Establish steady-state conditions: run the noise for at least 3 × expected RT60 (e.g., for an expected RT60 of 1.0 s, run for at least 3 seconds).
4. Abruptly switch off the noise source.
5. Record the decay at the receiver position using the SLM's RT60 module, which automatically fits a regression line to the decay curve and extrapolates to 60 dB.
6. Repeat at least 3 times at each receiver position to average out statistical fluctuation.
7. Move the receiver to the next position and repeat.

Disadvantages: Noisy (may disturb occupants in adjacent rooms). Requires substantial equipment. Multiple decay measurements needed at each position for averaging.

Method 2: Integrated Impulse Response (Swept Sine / Schroeder)

Procedure:

1. Set up the sound source and measurement microphone (or SLM) at the required positions.
2. Generate an exponential sine sweep (ESS) from 20 Hz to 20 kHz, typically 5–15 seconds long. The sweep is played through the loudspeaker and recorded by the microphone.
3. The recorded signal is deconvolved (cross-correlated with the inverse sweep filter) to extract the room impulse response (RIR).
4. The RIR is squared and backward-integrated (Schroeder method) to produce the energy decay curve.
5. T20 and T30 are derived from the linear regression of the decay curve over the -5 to -25 dB and -5 to -35 dB ranges respectively.

Disadvantages: Requires a laptop with measurement software (DIRAC by Brüel & Kjær, REW by John Mulcahy [free], EASERA by AFMG, AcousPlan's mobile measurement tool). Post-processing required. The swept sine method is sensitive to non-stationary noise (a door closing during the sweep corrupts the measurement).

Measurement Positions per ISO 3382-2

Receiver height: 1.2 m above floor (seated ear height) for rooms with seated audiences. 1.5 m for standing use (corridors, lobbies). The SLM or microphone should be on a tripod — never hand-held (body reflections corrupt high-frequency measurements).

Source height: 1.5 m above floor (standing speaker height) unless the room's typical use involves a different source height (e.g., 1.0 m for a PA loudspeaker mounted on a wall).

Data Processing

For each receiver position, calculate T20 and T30 at each octave band (125, 250, 500, 1000, 2000, 4000 Hz). Check the linearity of the decay curve: if the regression coefficient r² < 0.95, the measurement is unreliable (non-diffuse field, coupled spaces, or insufficient dynamic range).

Calculate the spatial average across all receiver positions:

RT60_avg = (1/N) × Σ RT60_i

where N is the number of receiver positions. Report the standard deviation to quantify spatial variation. ISO 3382-2 §6 specifies that spatial variation (coefficient of variation) should be less than 10% for a diffuse field. Higher variation indicates non-uniform absorption distribution or room geometry effects.

For compliance with standards that use a mid-frequency average (BB93, ANSI S12.60, WELL v2 Feature 74): RT60_mid = (RT60_500 + RT60_1000 + RT60_2000) / 3.

Part 3: Measuring STI (STIPA) per IEC 60268-16:2020

Equipment

1. STIPA signal source: A calibrated loudspeaker playing the STIPA test signal (IEC 60268-16 Annex D). The signal is a modulated pink noise containing two modulation frequencies per octave band. Pre-recorded WAV files are available from NTi Audio and Bedrock.
2. STIPA analyser: An SLM with STIPA analysis capability (NTi Audio XL2, Brüel & Kjær 2270 with BZ-7413 module, or Bedrock AM100).

Procedure

1. Place the loudspeaker at the typical talker position. Height: 1.5 m (standing talker) or 1.2 m (seated talker, e.g., for a meeting room).
2. Calibrate the source level to 60 dBA at 1 m (male normal speech level per IEC 60268-16 Table E.1). For female speech, use 58 dBA at 1 m.
3. Ensure background noise conditions are representative (HVAC running, no occupants unless measuring occupied STI).
4. Place the STIPA analyser at the first receiver position (1.2 m height, seated ear).
5. Start the STIPA analysis. Allow a minimum 15-second integration time for stable results.
6. Record the STI value. Most analysers also display octave-band MTI values — record these for diagnostic purposes.
7. Repeat at a minimum of 3 receiver positions per room.
8. Calculate the spatial average and minimum STI. The minimum STI (worst position) is the critical value for compliance — a room is only as good as its worst seat.

Common Pitfalls

Pitfall 1: Wrong source level. Setting the source to 70 dBA at 1 m artificially inflates the SNR, producing STI values 0.05–0.10 higher than real speech would achieve. Always calibrate to 60 dBA at 1 m.

Pitfall 2: Measurement too short. A 5-second STIPA integration produces results with standard deviation of ±0.05. A 15-second integration reduces this to ±0.02. Always use at least 15 seconds.

Pitfall 3: Background noise not representative. Measuring STI on a Sunday afternoon when HVAC is off, then reporting the result as the design-condition STI. Background noise is a direct input to the MTF calculation — it must be measured in the same condition that occupants will experience.

Pitfall 4: Source not directional enough. Using a highly directional loudspeaker (instead of omnidirectional) produces artificially high STI on-axis and artificially low STI off-axis. The source must approximate human speech directivity — dodecahedral sources are slightly too omnidirectional (real talkers are directional above 2 kHz), but they are the accepted standard.

Part 4: Measuring Background Noise per ISO 1996-2:2017

The Parameter

Background noise is typically quantified as LAeq,T (A-weighted equivalent continuous sound pressure level over measurement period T) in decibels. For octave-band analysis (needed for NC/NR curve assessment per ASHRAE/BS 8233), levels at 63, 125, 250, 500, 1000, 2000, 4000, and 8000 Hz are required.

Procedure

1. Set up the SLM at the receiver position (1.2 m height, at least 1 m from any surface, away from air supply diffusers and return air grilles unless measuring their contribution specifically).
2. Ensure the room is in its normal operating condition: HVAC running, lighting on, typical equipment active, but no occupants and no speech.
3. Measure LAeq for a minimum of 5 minutes (longer in rooms with intermittent noise sources such as lift motors or traffic).
4. Simultaneously record octave-band Leq values for NC/NR curve plotting.
5. Record LAmax and LA90 to characterise the noise floor and peak events.
6. Repeat at 3+ representative positions in the room.

NC/NR Curve Assessment

Plot the octave-band Leq values on the NC (Noise Criteria) or NR (Noise Rating) curve chart. The NC/NR rating of the room is the highest curve that is touched or exceeded by the measured spectrum at any octave band. For example, if the measured spectrum touches the NC 35 curve at 500 Hz but is below NC 35 at all other frequencies, the room rating is NC 35.

Part 5: Worked Example — Measuring a 200 m² Open-Plan Office

Project Brief

A 200 m² open-plan office (20 m × 10 m × 2.8 m) is undergoing WELL v2 Feature 74 certification. The acoustic consultant must verify three preconditions: P1 (background noise ≤ NC 40), P3 (RT60 ≤ 0.60 s), and the optimisation O8 (D₂,S ≥ 7 dB/dd, rD ≤ 5 m).

Equipment Used

• NTi Audio XL2 Sound Level Meter (Class 1, RT60 + STIPA modules)
• NTi Audio DS3 dodecahedral loudspeaker + NTi PA3 power amplifier
• NTi Audio MA1 measurement microphone (1/2" condenser, free-field, Class 1)
• Brüel & Kjær Type 4231 calibrator (94 dB at 1000 Hz)
• STIPA test signal WAV file (NTi Audio, IEC 60268-16:2020 Annex D compliant)
• Laptop with NTi Audio Report software for post-processing

Measurement Session (3 hours)

09:00 — Calibration check. SLM shows 94.0 dB at 1000 Hz. Within ±0.3 dB of target. Proceed.

09:10 — Background noise measurement.

Conditions: Normal business day, HVAC in automatic mode, no occupants (measured before staff arrival). 6 receiver positions across the floor plate, 5-minute LAeq at each.

Results (spatial average):

Octave-band analysis: highest NC curve touched is NC 37 (at 250 Hz). Result: NC 37 — passes WELL P1 (NC 40).

09:50 — RT60 measurement (interrupted noise method).

Source position 1: centre of room, 1.5 m height. Pink noise at 95 dBA at 1 m (background noise + 45 dB minimum headroom at each octave band verified). 6 receiver positions, 3 decays per position.

Results (spatial average, T30):

Mid-frequency average: (0.52 + 0.48 + 0.44) / 3 = 0.48 seconds — passes WELL P3 (≤ 0.60 s).

Note: the 125 Hz RT60 (0.92 s) is nearly twice the mid-frequency value — a typical low-frequency reverberation issue. If this were a DIN 18041 project, the bass ratio T₁₂₅/T₅₀₀ = 0.92/0.52 = 1.77 would fail the ≤ 1.2 requirement.

10:40 — STIPA measurement.

Source at typical talker position (standing presenter, 1.5 m). 60 dBA at 1 m. 6 receiver positions at desk height (1.2 m).

Results:

rD (STI = 0.50 distance): interpolated ≈ 7.5 m. Fails WELL O8 target (rD ≤ 5 m). The spatial decay D₂,S calculated from the STI vs distance data is approximately 4.8 dB/dd. Fails WELL O8 target (D₂,S ≥ 7 dB/dd).

Recommendation: Install freestanding screens (1.4 m height) between workstation clusters and sound masking system at 42 dBA to reduce rD to ≤ 5 m.

11:30 — Post-session calibration check. SLM reads 93.9 dB at 1000 Hz. Drift of -0.1 dB. Within ±0.5 dB tolerance. All measurements valid.

Report Contents

The acoustic measurement report should include:

1. Project details, room identification, and measurement date
2. Equipment list with serial numbers and calibration certificates
3. Measurement conditions (temperature, humidity, HVAC status, occupancy)
4. Measurement positions (floor plan with numbered source and receiver locations)
5. Results tables (per position and spatial average, at each octave band)
6. Compliance assessment against the applicable standard(s)
7. Pre- and post-session calibration check results
8. Photographs of the room showing measurement setup
9. Uncertainty statement per ISO 3382-2 §6.3
10. Recommendations for any non-compliant parameters

Related Reading:

• RT60 Measurement Methods Compared — interrupted noise vs impulse response vs swept sine
• Speech Transmission Index: The Complete Technical Reference — STI theory, calculation, and measurement
• RT60 Complete Reference — T20, T30, EDT definitions and optimal values