How to Measure Room Acoustics: Equipment, Methods, and What the Numbers Mean

Think of measuring a room's acoustics the way a doctor takes a patient's vital signs. You can press two fingers to someone's wrist and count heartbeats — it will tell you something useful, and it costs nothing. Or you can wire the patient to a twelve-lead ECG, capture waveform data across the entire cardiac cycle, and diagnose conditions that a pulse check would never reveal. Both are measurements. One is a rough check; the other is diagnostic-grade evidence. The equipment, the method, and the interpretation all determine whether you end up with an answer or an opinion.

Room acoustics works the same way. You can clap your hands in a room and listen to the decay — that is the two-finger pulse check. Or you can set up an omnidirectional loudspeaker, a calibrated microphone, and a swept sine signal, capture the room's impulse response at six positions, and extract reverberation time, early decay time, clarity, definition, and speech intelligibility to the precision that ISO 3382 demands. Both tell you something about the room. Only one will stand up to a compliance audit.

You have calculated the RT60. You have specified the treatment. The acoustic ceiling tiles are installed, the wall panels are mounted, the carpet is laid. Now you need to prove that it works — that the room actually meets BB93, WELL v2, DIN 18041, or whichever standard the project requires. That means measuring. Here is how to do it properly, what equipment you need at each level of rigor, and what the numbers actually mean once you have them.

What You Are Actually Measuring

When you measure a room's acoustics, you are capturing something called the room impulse response — abbreviated RIR. The impulse response is the complete acoustic signature of the room from a specific source position to a specific receiver position. It is a time-domain recording of what happens when a theoretically perfect, infinitely short burst of sound energy is released into the room.

In practice, no source produces a perfect impulse. But using techniques like deconvolution (which we will cover shortly), you can extract the impulse response from real-world signals with extraordinary precision.

Why the Impulse Response Matters

The impulse response contains all the acoustic information about the path between source and receiver. From a single RIR, you can extract every parameter defined in ISO 3382-1:2009 and ISO 3382-2:2008:

• RT60 (T20, T30) — reverberation time, the time for sound to decay by 60 dB
• EDT — early decay time, measured over the first 10 dB of decay (often more perceptually relevant than RT60)
• C80 — clarity, the ratio of early energy (0–80 ms) to late energy (after 80 ms), critical for music
• C50 — clarity for speech, the ratio of early energy (0–50 ms) to late energy
• D50 — definition, the fraction of total energy arriving within the first 50 ms
• STI — speech transmission index, derived from the modulation transfer function
• Ts — centre time, the time of the "centre of gravity" of the impulse response
• LF — lateral fraction, the proportion of early lateral energy (requires a figure-of-eight microphone)

Equipment Levels: From Free to Forensic

Not every project needs a Class 1 sound level meter and a dodecahedron speaker. The right equipment depends on the question you are trying to answer and the evidentiary standard your answer needs to meet.

Each level answers a different question. A smartphone balloon-pop measurement answers: "Is there an obvious problem?" A full ISO 3382-2 measurement answers: "Does this room comply with the specified standard, and here is the certified evidence to prove it."

What Makes Equipment ISO-Compliant

ISO 3382-2:2008 specifies requirements for both source and receiver. The sound source must be approximately omnidirectional — meaning it radiates sound energy equally in all directions. A dodecahedron speaker, which has twelve drivers arranged on the faces of a regular dodecahedron, is the standard solution. A single loudspeaker, regardless of quality, is directional by nature and will bias the measurement toward the direction the speaker is facing.

The microphone must also be omnidirectional. Class 1 sound level meters (per IEC 61672-1) have calibrated omnidirectional capsules with known frequency response, and the calibration is traceable to national standards. A USB measurement microphone like the UMIK-1 has a reasonably flat frequency response, but it is not calibrated to Class 1 precision and its calibration is not independently traceable. For compliance measurements that will be submitted to a WELL assessor or building control officer, the Class 1 instrument is not optional.

ISO 3382-2:2008 — The Measurement Standard for Ordinary Rooms

ISO 3382-2 is the reference standard for measuring reverberation time in ordinary rooms — offices, classrooms, hospitals, residential spaces, and any other room that is not a performance venue (performance spaces fall under ISO 3382-1). Understanding its requirements is essential if your measurements need to be defensible.

Source Requirements

The sound source must produce a sound pressure level at least 45 dB above the background noise level in each octave band from 125 Hz to 4000 Hz. This is a surprisingly demanding requirement. In a quiet office with a background noise level of 35 dBA, the source needs to produce at least 80 dB at the measurement position — which means considerably more at the source position, given distance attenuation. A balloon pop in a quiet room typically meets this. A handclap often does not, particularly at 125 Hz.

The source should be as close to omnidirectional as possible. ISO 3382-2 does not mandate a dodecahedron, but it notes that deviations from omnidirectionality will increase measurement uncertainty. In practice, a dodecahedron is the professional standard.

Microphone Requirements and Placement

The microphone must be omnidirectional and placed:

• At least 1 metre from any reflecting surface (walls, columns, furniture)
• At least 1 metre from the sound source
• At a height of 1.2 metres from the floor — this corresponds to the ear height of a seated listener, which is the primary use case for most rooms covered by ISO 3382-2

Minimum Number of Measurements

ISO 3382-2 requires a minimum of 6 source-receiver combinations: at least 2 source positions and at least 3 receiver positions per source. This is not arbitrary — it captures the spatial variation of reverberation time within the room.

Reverberation time is not a single number that applies uniformly across a room. In a rectangular conference room, RT60 measured near a corner can differ from RT60 measured at the room centre by 0.1 to 0.3 seconds, depending on the room's absorption distribution. A single measurement at one position can be misleading. Six measurements, averaged, produce a result with known uncertainty.

For the "survey" method (reduced accuracy), a minimum of 1 source position and 2 receiver positions is permitted, but the measurement uncertainty increases. For compliance with WELL v2 or BB93, the full 6-position method is strongly recommended.

Conditions During Measurement

The room must be furnished but unoccupied. Furniture contributes absorption — particularly upholstered chairs, which can add 0.3 to 0.5 m² Sabine per chair. Measuring an unfurnished room will yield higher RT60 values than the room will have in use, producing a misleadingly pessimistic result.

Windows must be closed. Open windows change both the room volume (technically) and the absorption (significantly — an open window acts as a perfect absorber with alpha = 1.0).

HVAC must be running at normal operating levels. The HVAC system contributes background noise that affects the measurable decay range, and in some systems, air movement can affect the sound field. The standard requires that the measurement reflects the room as it will actually be used.

Three Methods for Capturing the Impulse Response

There are three established methods for measuring a room's impulse response. Each has trade-offs between convenience, repeatability, and signal quality.

Method 1: Interrupted Noise

The oldest method. You fill the room with broadband noise from a loudspeaker, then abruptly switch it off and record the decay. The shape of the decay curve directly gives the reverberation time.

Advantages: Simple to execute. No post-processing required — the decay is directly visible in the recorded signal. Well-understood and referenced in the original ISO 354:2003 method for reverberation room measurements.

Disadvantages: Poor signal-to-noise ratio. The tail of the decay curve (the quiet end, which determines T30 accuracy) is easily masked by background noise. Each measurement captures only one decay, so you need to repeat at least three times per position and average. The method does not produce a clean impulse response, which means you cannot extract parameters like C80, D50, or STI from the data. You only get RT60.

Best for: Quick RT60 checks in rooms where RT60 is the only parameter of interest and background noise is low.

Method 2: Impulsive Source (Balloon Pop or Starter Pistol)

An impulsive source — a balloon pop, a starter pistol blank, or a purpose-built impulse generator — creates a short burst of broadband sound energy. The microphone captures the direct sound followed by the room's reflections and decay. This recording is, approximately, the room's impulse response.

Advantages: Very fast. No computer or playback equipment needed in the room. A balloon and a microphone are all you carry.

Disadvantages: Repeatability is poor. The spectrum of a balloon pop depends on the balloon diameter, inflation pressure, and how it is burst. A small party balloon produces adequate energy above 500 Hz but very little at 125 Hz. A large weather balloon (60 cm diameter) produces better low-frequency energy but is cumbersome. Starter pistols produce very high peak levels that can overdrive microphone preamps if gain is not carefully set. Because the source is impulsive, you get one shot per measurement — if the recording clips or the background noise intrudes, you repeat the entire measurement.

ISO compliance: Partially compliant. ISO 3382-2 permits impulsive sources but notes that "the reproducibility of the excitation signal should be verified." In practice, this means you need to demonstrate consistent spectral content across your measurements, which is difficult with balloons.

Best for: Site surveys, preliminary assessments, and situations where you need data quickly and the measurement does not need to be ISO-certified.

Method 3: Swept Sine (ESS) or Maximum Length Sequence (MLS)

This is the professional standard. A computer generates a known signal — either an exponential sine sweep (ESS, also called a logarithmic swept sine) or a Maximum Length Sequence (MLS, a pseudo-random binary signal). The signal is played through the loudspeaker and simultaneously recorded by the microphone. In post-processing, the recorded signal is deconvolved against the original signal to extract the room impulse response.

Advantages: Best signal-to-noise ratio of any method. The swept sine technique can achieve 80–100 dB of dynamic range — far more than the 45 dB minimum required by ISO 3382-2. Because the excitation signal is known exactly, the deconvolution process rejects uncorrelated noise (traffic, HVAC, footsteps in the corridor), producing a clean impulse response even in moderately noisy environments. The ESS method also separates harmonic distortion from the linear impulse response, which MLS cannot do.

Disadvantages: Requires a computer (or dedicated instrument) to generate the signal and perform the deconvolution. The measurement takes longer — a typical sine sweep runs 5 to 15 seconds, compared to the fraction of a second for a balloon pop. Equipment setup is more involved.

ISO compliance: Fully compliant. ESS is the method recommended by the IEC for impulse response measurement and is the default in professional tools like Dirac (Bruel & Kjaer) and EASERA (AFMG).

Best for: All formal measurements, compliance assessments, research, and any situation where you need C80, D50, STI, or other parameters beyond RT60.

Extracting Parameters: Schroeder Backward Integration

Once you have the room impulse response, you need to extract the reverberation time from it. The standard method is Schroeder backward integration, published by Manfred Schroeder in 1965 and formalized in ISO 3382-1:2009.

The Problem with Forward Measurement

If you simply look at the impulse response waveform and try to find where the level has dropped by 60 dB, you will get an unreliable answer. The impulse response is a noisy, fluctuating signal — individual reflections create peaks and valleys that make it impossible to define a smooth decay visually. You would need to average many impulse responses to get a usable decay curve, which is exactly what the interrupted noise method does (and why it requires at least three repetitions).

The Schroeder Solution

Schroeder showed that you can obtain the equivalent of an infinite number of averaged interrupted-noise decays from a single impulse response by integrating the squared impulse response backwards in time:

E(t) = integral from t to infinity of h-squared(tau) d-tau

Where h(t) is the impulse response and E(t) is the energy decay curve. Plotting 10 log10(E(t) / E(0)) in decibels gives a smooth, monotonically decreasing curve — the Schroeder decay curve.

T20 and T30: Why We Do Not Measure the Full 60 dB

In theory, RT60 requires a 60 dB decay. In practice, achieving 60 dB of usable dynamic range in a real room is extremely difficult. Background noise in a typical office is 30–40 dBA. The source level at the microphone might be 80–90 dB. That gives a dynamic range of 40–60 dB — often not enough for a direct 60 dB measurement, especially at low frequencies where background noise is higher and source energy is lower.

ISO 3382-2 therefore defines two standard evaluation ranges:

• T20: measured over the decay from -5 dB to -25 dB (a 20 dB range), then extrapolated to 60 dB by multiplying by 3
• T30: measured over the decay from -5 dB to -35 dB (a 30 dB range), then extrapolated to 60 dB by multiplying by 2

Why extrapolation works: In a room with a single exponential decay (which is the assumption underlying all statistical reverberation formulas), the decay rate is constant. Measuring 20 dB of decay and multiplying by 3 gives the same answer as measuring 60 dB directly. In rooms with coupled volumes or non-exponential decay (auditoriums with stage houses, for example), T20 and T30 will differ — and that difference itself is diagnostically useful, indicating that the room has a double-slope decay.

Common Measurement Mistakes

Even with good equipment, measurement errors can invalidate results. These are the mistakes that turn up most frequently in practice.

1. Measuring with People in the Room

People absorb sound. An adult human body has an absorption area of approximately 0.40 to 0.55 m² Sabine, depending on clothing. In a 50 m² meeting room with 10 people, the human absorption adds 4 to 5.5 m² Sabine to the room's total. In a moderately treated room with total absorption around 25 m² Sabine, that is a 16–22% increase in absorption — enough to reduce RT60 by 0.1 to 0.2 seconds. ISO 3382-2 requires measurements in unoccupied rooms specifically because of this effect.

2. Too Few Measurement Positions

RT60 varies spatially. Near a wall corner, where three reflective surfaces converge, the local sound field is reinforced and the measured RT60 tends to be shorter. In the centre of the room, RT60 tends to be longer. Measuring at a single position can give a result that differs from the spatial average by 0.1 to 0.2 seconds in a typical room, and by more in rooms with strongly non-uniform absorption. The ISO minimum of 6 source-receiver combinations is designed to capture this variance and produce a meaningful average.

3. Background Noise Too High

If the background noise level is within 35 dB of the source level at the measurement position, you cannot reliably measure T30. If it is within 20 dB, you cannot reliably measure T20 either. The decay curve will flatten out at the noise floor, and the regression line used to calculate the decay slope will be biased. The most common cause: measuring during working hours with HVAC at full load and traffic noise entering through facade glazing. Solution: measure early morning, evening, or weekends when background noise is lowest — but keep HVAC running at its normal daytime setting if ISO 3382-2 conditions are required.

4. Source Not Omnidirectional

A directional loudspeaker sends more energy in some directions than others. Surfaces directly in front of the speaker receive more early energy; surfaces behind it receive less. The resulting impulse response is biased by the speaker's radiation pattern, and the extracted RT60 may differ from what an omnidirectional source would produce. This is particularly problematic for parameters like C80 and D50, which are sensitive to the balance between early and late energy. If you must use a non-omnidirectional source, rotate it to multiple orientations and average, or use it only for RT60 (which is less sensitive to source directivity than clarity metrics).

5. Not Reporting Octave-Band Results

A single-number RT60 — "the room has an RT60 of 0.6 seconds" — hides critical frequency-dependent information. A room might have an RT60 of 0.4 seconds at 2000 Hz and 1.1 seconds at 125 Hz. The single-number average might look compliant, but the low-frequency excess is a real problem that will manifest as boom, muddiness, and poor speech clarity for male voices.

ISO 3382-2 requires reporting RT60 at each octave band from 125 Hz to 4000 Hz, plus standard deviation. Standards like BB93 and ANSI S12.60 set frequency-dependent limits. A measurement that reports only a single number cannot be used for compliance verification.

6. Not Calibrating the Microphone

If you are using the measurement to report absolute sound pressure levels (for background noise assessment, for example), an uncalibrated microphone introduces unknown systematic error. Calibration should be performed at the start and end of each measurement session using a pistonphone or acoustic calibrator (94 dB or 114 dB at 1 kHz). The difference between start and end calibrations should be less than 0.5 dB; if it is larger, the measurement session should be repeated.

Smartphone Measurements: Useful but Not Sufficient

Smartphone apps for acoustic measurement have improved dramatically. AcousPlan's mobile measurement feature, along with apps like Studio Six Digital's AudioTools and Faber Acoustical's SoundMeter, can capture decay curves and estimate RT60 from a balloon pop or handclap recorded through the phone's built-in microphone.

What Smartphone Measurements Can Do

• Quick site surveys: Walk through a building, clap in each room, and get an approximate RT60 in 30 seconds. This is valuable for identifying problem rooms that need detailed assessment.
• Before-and-after comparisons: Measure before treatment is installed, then measure after. Even if the absolute values are imprecise, the relative change is meaningful if you use the same phone, same position, and same technique.
• Early design validation: If your prediction says RT60 will be 0.7 seconds and your smartphone measurement shows 1.3 seconds, something is wrong — and you did not need a Class 1 meter to detect that.
• Client communication: Showing a client a real-time decay curve on a phone screen is more persuasive than a spreadsheet. It builds understanding and trust.

What Smartphone Measurements Cannot Do

• Meet ISO 3382-2 requirements: The microphone is not omnidirectional (phone microphones are typically cardioid or hypercardioid), not calibrated to known standards, and has unknown frequency response characteristics, particularly below 200 Hz where many phones have high-pass filters. No WELL assessor or building control officer will accept a smartphone measurement as compliance evidence.
• Provide reliable absolute levels: Without calibration, a phone measurement might report "45 dBA" when the actual level is 38 or 52 dBA. The error is unpredictable and varies between phone models.
• Capture low-frequency performance accurately: Most smartphone microphones have poor sensitivity below 200 Hz. The 125 Hz octave band — critical for detecting bass buildup problems — is essentially unmeasurable with a phone.
• Produce defensible STI values: STI requires accurate modulation transfer function measurement, which demands known source characteristics and calibrated signal-to-noise ratios. A smartphone cannot provide this.

The Bottom Line on Phones

Smartphone measurements are the pulse check. They are valuable, they are free, and every acoustician should use them for rapid assessment. But they are not the ECG. When the project requires certified compliance — when a WELL credit, a BB93 approval, or a contract sign-off depends on the numbers — professional equipment and ISO 3382-2 methodology are non-negotiable.

What to Do with the Results

You have measured the room. You have impulse responses from 6 positions, extracted T20 or T30 at each octave band, and calculated the spatial average with standard deviation. Now what?

Compare Against Design Predictions

The first check is: does the measured RT60 match the prediction you made during design? If you used the Sabine or Eyring equation with manufacturer-supplied absorption coefficients, how close is the prediction to reality?

A well-executed prediction should be within 10–15% of the measured value. If the discrepancy is larger, investigate:

• Were the absorption coefficients accurate? Manufacturer data is measured per ISO 354 in a reverberation chamber — laboratory conditions that do not perfectly replicate in-situ performance. Mounting conditions matter. A ceiling tile mounted directly to the slab (Type A mounting) performs differently from the same tile in a suspended grid with a 200 mm plenum (Type E-200 mounting).
• Were all absorption sources accounted for? Furniture, people's belongings, equipment, and air gaps around doors all contribute absorption that is easy to omit from a calculation.
• Is the room geometry as built? Columns, soffits, and services that were not in the design model change the room volume and surface areas.

Check Compliance with Target Standard

Apply the standard that governs your project:

• BB93:2015 (UK schools): RT60 limits by room type and volume, measured furnished and unoccupied. Primary classrooms under 250 m³: maximum 0.6 s mid-frequency average.
• ANSI S12.60-2010 (US schools): RT60 maximum 0.6 s for core learning spaces under 283 m³ (10,000 ft³), plus background noise limit of 35 dBA.
• WELL v2 Sound concept: Feature S07 specifies RT60 limits by room type and volume. Verification requires measurement per ISO 3382-2.
• DIN 18041:2016 (German rooms): RT60 targets by room type and usage, with frequency-dependent limits.

Identify Problem Frequencies

Look at the octave-band profile. In well-designed rooms, RT60 decreases gradually from low to high frequencies (because most absorptive materials are more effective at higher frequencies). Red flags include:

• RT60 at 125 Hz more than twice the mid-frequency average: This indicates a bass buildup problem. The room lacks low-frequency absorption, which is common because most standard acoustic products (ceiling tiles, thin wall panels) have poor performance below 250 Hz. Solutions include membrane absorbers, Helmholtz resonators, or thick (100 mm+) porous absorbers with an air gap.
• RT60 rising at 4000 Hz: Unusual, and typically indicates a measurement problem (background noise contamination at high frequencies) rather than a room problem. Check the signal-to-noise ratio at 4000 Hz.
• Large standard deviation at specific frequencies: Spatial variation is frequency-dependent. Large variance at low frequencies is expected (room modes cause position-dependent behavior). Large variance at mid-frequencies suggests non-uniform absorption distribution — one end of the room may be significantly more absorptive than the other.

Verify Treatment Effectiveness

If you are measuring after acoustic treatment has been installed, compare the post-treatment results with the pre-treatment measurements (or the untreated prediction). Calculate the reduction in RT60 at each octave band and compare it with the expected reduction based on the treatment manufacturer's absorption data.

If the treatment is underperforming, common causes include:

• Insufficient coverage: The treatment covers less surface area than specified in the acoustic design.
• Wrong mounting: Panels mounted flat against a wall perform differently from panels mounted with an air gap. The air gap increases low-frequency absorption but may reduce high-frequency performance.
• Wrong product: Substitutions happen on construction sites. A panel rated NRC 0.85 was specified; NRC 0.55 was delivered and installed without the acoustician being notified. Always verify the product on site before measuring.

Pulling It All Together

Acoustic measurement is not difficult, but it is methodical. The quality of your results depends entirely on whether you followed the method, and whether your equipment was appropriate for the precision you need.

For design validation and quick checks, a smartphone app and a balloon are genuinely useful tools. Do not underestimate them. They will catch gross errors and save you from discovering problems at handover.

For compliance verification, there is no shortcut. ISO 3382-2 exists for a reason: it defines the minimum methodology that produces results with known, bounded uncertainty. The equipment costs money. The method takes time. But when a WELL credit or a building approval depends on the numbers, that investment is trivial compared to the cost of remediation if the room fails.

And if you are at the stage where you have not built the room yet — where you are still in design, still choosing materials, still deciding whether you need acoustic treatment at all — start with prediction. Model the room. Calculate the RT60 at each octave band using Eyring (not Sabine, unless the room is a reverberant hall). Check the compliance target. Specify the treatment. Then build. Then measure. The measurement confirms the design, or it tells you what to fix. Either way, you need it.

Try It Yourself

Use AcousPlan's acoustic calculator to predict RT60 for your room before you measure. Enter your room dimensions and surface materials, and the tool will compute reverberation time at each octave band using both Sabine and Eyring formulas — so you know what your measurements should show before you pick up a microphone.

If you have already measured, AcousPlan's measurement import feature lets you upload your RT60 data and overlay it against the predicted values. You can see exactly where the prediction and the measurement agree, where they diverge, and what that divergence means for your design.

References

• ISO 3382-1:2009 — Acoustics — Measurement of room acoustic parameters — Part 1: Performance spaces
• ISO 3382-2:2008 — Acoustics — Measurement of room acoustic parameters — Part 2: Reverberation time in ordinary rooms
• ISO 3382-3:2012 — Acoustics — Measurement of room acoustic parameters — Part 3: Open plan offices
• ISO 354:2003 — Acoustics — Measurement of sound absorption in a reverberation room
• IEC 60268-16:2020 — Sound system equipment — Part 16: Objective rating of speech intelligibility by speech transmission index
• IEC 61672-1:2013 — Electroacoustics — Sound level meters — Part 1: Specifications
• ANSI/ASA S12.60-2010 — Acoustical Performance Criteria, Design Requirements, and Guidelines for Schools
• BB93:2015 — Acoustic design of schools: performance standards (UK Department for Education)
• DIN 18041:2016 — Acoustic quality in rooms — Specifications and instructions for the room acoustic design
• WELL v2 Sound concept — Feature S07: Sound Mapping
• Schroeder, M. R. (1965). "New Method of Measuring Reverberation Time." Journal of the Acoustical Society of America, 37(3), 409–412.