WELL v2 Feature 74 Decoded: Every Acoustic Requirement, Every Calculation, Every Clause

The Standard That Trips Up Experienced Assessors

WELL Building Standard v2, administered by the International WELL Building Institute (IWBI), reorganized its acoustic requirements under the Sound concept beginning with Q3 2020. What was previously a single feature — Feature 74 (Sound) in WELL v1 — has been restructured into a set of features under the Sound concept (S01 through S07), but the original Feature 74 numbering and its three-part structure remain the most widely referenced framework among practitioners and WELL assessors preparing for certification.

The three parts of Feature 74 address fundamentally different acoustic phenomena:

• Part 1 targets reverberation and background noise in enclosed rooms (meeting rooms, private offices, telephone rooms).
• Part 2 targets ambient noise levels in open plan environments.
• Part 3 targets speech privacy — and this is where most projects fail.

This article is the comprehensive reference. Every requirement, every threshold, every calculation, in one place.

Part 1: Sound-Isolating Rooms — Reverberation and Background Noise

Part 1 applies to enclosed spaces where speech intelligibility and acoustic comfort are essential: meeting rooms, private offices, telephone rooms, and focus rooms. The requirements are:

Reverberation Time (RT60)

All enclosed rooms listed below must achieve an RT60 equal to or less than the specified target. Measurements must be taken in the furnished, occupied condition (or furnished with representative absorption to simulate occupancy). The measurement methodology is ISO 3382-2:2008, which specifies the interrupted noise method or the integrated impulse response method for ordinary rooms.

The distinction between Sabine (§A.1) and Eyring (§A.2) matters for larger rooms. Sabine's formula assumes diffuse field conditions with relatively low average absorption coefficients (< 0.3). For large conference rooms with substantial acoustic treatment, Eyring's formula provides more accurate predictions. In practice, WELL assessors should use whichever method yields the more conservative (higher) RT60 estimate and verify that it still falls within the threshold.

Key detail: The "furnished, occupied condition" requirement means that measuring an empty room during commissioning will typically yield an RT60 that is 0.1-0.3 seconds longer than the operational condition. People absorb sound — a seated adult in a meeting room contributes roughly 0.4-0.5 m² Sabine of absorption. A meeting room designed for 12 people gains approximately 5-6 m² Sabine of absorption just from the occupants. This is why some projects pass during a populated demonstration but fail when measured in the unoccupied state that some assessors incorrectly apply.

Background Noise Levels

Background noise in Part 1 spaces must not exceed 35 dBA (L_Aeq measured over a representative period, typically 1 hour during normal HVAC operation). This aligns with ASHRAE noise criteria NC-30 to NC-35, which is achievable in most modern HVAC designs but requires attention to:

• Duct velocity: Keep below 5 m/s in final branches serving meeting rooms
• Diffuser selection: Slot diffusers and perforated face diffusers are quieter than drum louvers
• VAV box placement: Minimum 3 m of lined ductwork between VAV box and nearest diffuser
• Breakout noise: Rectangular duct near meeting rooms should be internally lined or wrapped

Part 2: Open Plan Noise Levels — The Background Noise Balance

Part 2 addresses open plan work areas, which present a fundamentally different acoustic challenge. In enclosed rooms, the goal is quiet. In open plan environments, the goal is controlled noise — enough background sound to mask speech from neighboring workstations, but not so much that it becomes fatiguing.

Requirements

The measurement specification per IEC 61672-1 requires a Class 1 or Class 2 integrating-averaging sound level meter. Measurements are taken at desk height (1.2 m) at representative workstation positions — not just at the center of the room. At least five measurement positions should be used for spaces up to 500 m², with additional points for larger floor plates.

The HVAC Paradox

The two-tier noise requirement (overall ≤ 45 dBA, mechanical ≤ 35 dBA) creates a specific design tension. If HVAC noise alone is 35 dBA, and the overall target allows up to 45 dBA, there is a 10 dB "gap" that must be filled by either:

1. Occupant activity noise (speech, typing, movement) — uncontrollable and variable
2. Sound masking systems — controllable and uniform

HVAC Design for Compliance

Achieving the mechanical noise threshold of 35 dBA in an open plan space requires careful HVAC design. Common failures include:

• Undersized return air paths: Return air through ceiling plenum voids generates turbulent noise. Ducted returns are quieter but more expensive.
• Rooftop unit vibration: Structure-borne noise from rooftop packaged units transmits through the deck. Spring isolators with 95%+ isolation efficiency are essential.
• Fan coil units in the ceiling: FCUs located directly above workstations can produce 38-42 dBA at desk level. Select units rated NC-25 or lower, and locate them over circulation paths rather than desk clusters.
• VAV reheat box noise: Pressure-independent VAV boxes with fiberglass-lined attenuators reduce downstream noise by 10-15 dB.

Part 3: Speech Privacy — The Calculation Most Assessors Get Wrong

Part 3 is the compliance requirement that causes the most project failures. The requirement is deceptively simple to state:

The Speech Transmission Index (STI) between adjacent workstations, and between private offices and adjacent corridors, must be less than 0.50.

An STI of 0.50 corresponds to "fair" intelligibility on the IEC 60268-16 scale. Below 0.50, a listener can hear that someone is speaking but cannot reliably discern the words. This is the threshold for acceptable speech privacy in a workplace context.

Above 0.50, conversations become intelligible — distracting in open plan environments and a confidentiality risk for private offices.

The STI Calculation: IEC 60268-16 STIPA Method

The Speech Transmission Index quantifies how much of the speech signal's temporal modulation structure is preserved as it travels from a speaker to a listener through a room. The calculation proceeds through four stages.

Stage 1: Determine the Signal-to-Noise Ratio (SNR) per Octave Band

For each of the seven octave bands used in the STIPA method (125 Hz, 250 Hz, 500 Hz, 1000 Hz, 2000 Hz, 4000 Hz, 8000 Hz), calculate the effective signal-to-noise ratio at the receiver position:

SNR(f) = L_s(f) - L_n(f) [dB]

Where:

• L_s(f) is the speech level at the receiver in octave band f, accounting for distance attenuation, room absorption, and any barriers
• L_n(f) is the background noise level (including sound masking) in octave band f at the receiver

Speech level at distance d is calculated as:

L_s(f, d) = L_s(f, 1m) - 20 log10(d) - excess_attenuation(f)

Where excess attenuation includes the effect of room absorption on the reverberant field contribution.

Stage 2: Compute the Apparent Signal-to-Noise Ratio

The apparent SNR is bounded:

SNR_apparent(f) = max(-15, min(15, SNR(f))) [dB]

This clipping to the range [-15 dB, +15 dB] reflects the limits of auditory perception. At +15 dB SNR, speech is essentially perfectly intelligible; below -15 dB, it is completely unintelligible. Further improvements or degradations beyond these bounds do not affect perceived intelligibility.

Stage 3: Compute the Modulation Transfer Function (MTF)

The modulation transfer function for each octave band accounts for the effect of reverberation on temporal modulation. For a diffuse reverberant field:

m(f, F_mod) = 1 / sqrt(1 + (2 pi F_mod * T(f) / 13.8)²)

Where:

• F_mod is the modulation frequency (STIPA uses two modulation frequencies per octave band)
• T(f) is the RT60 in octave band f
• 13.8 is the decay constant derived from the Schroeder integral (60 dB / ln(10) ln(10) / (60 ln(10) / (60/13.8)))

m_eff(f, F_mod) = m(f, F_mod) * 10^(SNR_apparent(f)/10) / (1 + 10^(SNR_apparent(f)/10))

This equation shows the critical interaction: even if the SNR is favorable (high L_s, low L_n), a long RT60 can degrade the modulation transfer to the point where STI exceeds 0.50. Conversely, even moderate reverberation can be compensated by sufficient background noise (which reduces the effective SNR and thus the STI).

Stage 4: Weighted Average to STI

The STIPA method averages the modulation transfer indices across octave bands using the IEC 60268-16 weighting factors:

Note: The alpha weights apply to the modulation transfer indices within each band; the beta weights apply to redundancy corrections between adjacent bands. The final STI is:

STI = sum(alpha_k MTI_k) - sum(beta_j sqrt(MTI_j * MTI_{j+1}))

Where MTI_k is the modulation transfer index for band k, derived from the average of the effective modulation transfer values at the two STIPA modulation frequencies for that band.

An STI < 0.50 means the speech modulation structure is sufficiently degraded that listeners cannot reliably identify words. This is the WELL Feature 74 Part 3 pass criterion.

Full Worked Example: 400 m² Open Plan Office

Consider a typical open plan floor plate that many WELL projects encounter:

Room Parameters

• Floor area: 400 m² (20 m x 20 m)
• Ceiling height: 3.0 m
• Room volume: 1,200 m³
• Ceiling: Mineral fiber acoustic tile, NRC 0.70
• Floor: Carpet tile, NRC 0.30
• Walls: Glass curtain wall (2 sides), plasterboard (2 sides)
• Furniture: 80 workstations on 2.5 m x 2.5 m grid, 1.2 m high screens between desks

Step 1: Calculate RT60

Using Sabine's formula: RT60 = 0.161 V / A

Total absorption at 500 Hz:

• Ceiling: 400 m² x 0.70 = 280 m² Sabine
• Floor: 400 m² x 0.30 = 120 m² Sabine
• Glass walls: 2 x (20 m x 3.0 m) x 0.05 = 6 m² Sabine
• Plasterboard walls: 2 x (20 m x 3.0 m) x 0.10 = 12 m² Sabine
• Furniture and occupants: ~40 m² Sabine (estimated)
• Total A = 458 m² Sabine

This comfortably passes Part 1's RT60 threshold of 0.6 seconds. Good.

Step 2: Check Background Noise

Measured HVAC noise: 33 dBA. This passes Part 2's mechanical noise threshold of 35 dBA.

Without sound masking, total background noise during low-occupancy periods: approximately 34-36 dBA.

Step 3: Calculate STI Between Adjacent Workstations (The Critical Check)

Source: Speaker at workstation A, normal vocal effort. Receiver: Occupant at workstation B, 2.5 m away (adjacent desk).

Speech level at receiver (500 Hz octave band): L_s(500, 2.5m) = 61.2 - 20 * log10(2.5) - 0.5 (excess attenuation from screen) L_s(500, 2.5m) = 61.2 - 8.0 - 0.5 = 52.7 dB

Background noise at receiver (500 Hz, no masking): L_n(500) = 30 dB (HVAC contribution in 500 Hz band, typical for a 33 dBA overall system)

SNR(500) = 52.7 - 30 = 22.7 dB -> clipped to 15 dB (apparent SNR)

Modulation transfer (500 Hz, F_mod = 1.0 Hz): m(500, 1.0) = 1 / sqrt(1 + (2 pi 1.0 * 0.42 / 13.8)²) m(500, 1.0) = 1 / sqrt(1 + (0.191)²) m(500, 1.0) = 1 / sqrt(1.036) m(500, 1.0) = 0.982

Effective modulation transfer: m_eff(500, 1.0) = 0.982 * 10^(15/10) / (1 + 10^(15/10)) m_eff(500, 1.0) = 0.982 * 31.62 / 32.62 m_eff(500, 1.0) = 0.952

This extremely high modulation transfer value at 500 Hz — the most heavily weighted band — means that the speech modulation structure is almost perfectly preserved. Repeating this calculation across all seven octave bands and applying the STIPA weighting yields:

STI (no masking, 2.5 m distance) = 0.55

This FAILS WELL Feature 74 Part 3. The STI of 0.55 exceeds the 0.50 threshold.

The room passes Part 1 (RT60 = 0.42 s) and Part 2 (HVAC noise = 33 dBA). But speech between adjacent desks at 2.5 m is too intelligible. The occupant at workstation B can understand what the person at workstation A is saying. Speech privacy is not achieved.

Why Increasing Desk Spacing Alone Is Not Enough

If we increase workstation spacing to 3.5 m:

L_s(500, 3.5m) = 61.2 - 20 * log10(3.5) - 0.7 = 61.2 - 10.9 - 0.7 = 49.6 dB SNR(500) = 49.6 - 30 = 19.6 dB -> clipped to 15 dB

The SNR is still capped at +15 dB. The STI barely changes — perhaps dropping to 0.52-0.53. Still a fail. The problem is not distance. The problem is that the background noise is too low relative to the speech level.

The Fix: Sound Masking at 45 dBA

Install a sound masking system calibrated to 45 dBA overall (pink noise spectrum shaped to approximate speech masking). The 500 Hz octave band contribution of a 45 dBA masking system is approximately 40 dB.

Revised calculation at 2.5 m spacing with masking:

L_n(500) = 10 * log10(10^(30/10) + 10^(40/10)) = 40.4 dB (HVAC + masking combined)

SNR(500) = 52.7 - 40.4 = 12.3 dB (no longer clipped — this is within the -15 to +15 range)

Effective modulation transfer with masking: m_eff(500, 1.0) = 0.982 * 10^(12.3/10) / (1 + 10^(12.3/10)) m_eff(500, 1.0) = 0.982 * 17.0 / 18.0 m_eff(500, 1.0) = 0.927

Across all octave bands, the masking system has the largest impact in the 500-2000 Hz range — exactly the range that carries the most speech intelligibility information and receives the highest STIPA weighting.

STI (with 45 dBA masking, 2.5 m distance) = 0.42

This PASSES WELL Feature 74 Part 3. The STI of 0.42 is comfortably below the 0.50 threshold.

The sound masking system reduced the STI from 0.55 to 0.42 — a shift from "fair" to "poor" intelligibility, which in the context of speech privacy is exactly the desired outcome. Occupants at adjacent desks can hear that conversation is occurring but cannot reliably determine what is being said.

Common Failure Modes

Projects pursuing WELL certification frequently fail the acoustic requirements for predictable, avoidable reasons. The table below catalogues the most common failure modes observed across WELL assessment projects.

WELL Feature 74 Requirements by Space Type

The following table consolidates the complete set of acoustic thresholds across Part 1, Part 2, and Part 3 for each applicable space type.

Reading the table: "N/A" means that particular Part does not apply to that space type. For open plan workstations, Part 1 (enclosed room reverberation) does not apply because the space is not enclosed. For meeting rooms, Part 2 (open plan noise levels) and Part 3 (speech privacy between workstations) do not apply — though large meeting rooms should still aim for good speech intelligibility (STI > 0.60) for communication purposes, which is a separate consideration from privacy.

The Interaction Between RT60, Background Noise, and STI

Many practitioners treat RT60 and STI as independent parameters. They are not. The STI calculation explicitly incorporates both reverberation time (through the modulation transfer function) and background noise level (through the signal-to-noise ratio). Understanding their interaction is essential for efficient design.

Low RT60, Low BGN = High STI (Privacy Fail)

A well-absorbed room with low background noise preserves speech modulation extremely well. This is ideal for a meeting room (where you want high intelligibility) but disastrous for speech privacy in an open plan office. This is the scenario in our worked example above — RT60 of 0.42 s and BGN of 34 dBA produced an STI of 0.55, which fails Part 3.

High RT60, Low BGN = Moderate STI (May Pass or Fail)

A reverberant room degrades the modulation transfer function, which reduces STI. But high reverberation also means the reverberant speech energy fills the room, potentially maintaining speech levels at greater distances. The net effect depends on the specific geometry. In many cases, an RT60 of 0.7-0.8 s in an open plan space (achievable with a hard ceiling) may actually improve speech privacy — but at the cost of failing Part 1's RT60 threshold (if applicable) and causing general acoustic discomfort.

This is not a recommended strategy. Designing a reverberant room for privacy is acoustically inelegant and creates a host of secondary problems (increased Lombard effect, listener fatigue, reduced communication quality for intended conversations).

Low RT60, High BGN = Low STI (Privacy Pass)

The optimal open plan design: absorptive ceiling to control reverberation and improve spatial decay, combined with sound masking to raise the background noise floor. The masking reduces the effective SNR across all octave bands, degrading the modulation transfer and pushing STI below 0.50.

This combination — RT60 between 0.3-0.5 s plus sound masking at 42-45 dBA — is the reliable path to WELL Feature 74 compliance across all three Parts.

Measurement and Verification Protocol

WELL certification requires on-site acoustic measurement by a WELL Performance Testing Agent. The measurement protocol for Feature 74 compliance must include:

Equipment

• Class 1 integrating-averaging sound level meter per IEC 61672-1
• Omnidirectional sound source per ISO 3382-2 (for RT60 measurement)
• STIPA signal generator and analyzer (or software-based equivalent per IEC 60268-16)

Measurement Positions

• RT60: Minimum 3 source-receiver combinations per room, receiver positions at 1.2 m height, minimum 1.5 m from any surface, minimum 2 m from source
• Background noise: 5 positions per open plan zone at desk height (1.2 m), excluding positions within 2 m of supply air diffusers
• STI: Measured between representative workstation pairs at the actual desk-to-desk distance. Minimum 3 measurement pairs per open plan zone. Source at 1.5 m height (standing speaker), receiver at 1.2 m height (seated listener)

Conditions

• All HVAC systems operating at normal design condition
• Sound masking system (if installed) operating at calibrated level
• Doors to enclosed rooms closed during RT60 and BGN measurements
• Ambient conditions: no construction activity, no unusual external noise events

Reporting

Measurement reports must include: instrument calibration certificates, measurement positions on a floor plan, raw measurement data, calculated parameters (RT60, L_Aeq, STI), comparison against WELL thresholds, and pass/fail determination for each Part.

Design Guidance: Getting It Right the First Time

The cost of acoustic remediation after construction is 3-5 times the cost of incorporating the same treatments during the design phase. Here is the minimum specification to achieve WELL Feature 74 compliance for a typical open plan office:

1. Acoustic ceiling tile: NRC 0.85 or higher, full coverage. Budget: $25-40/m².
2. Carpet or carpet tile: NRC 0.25-0.35. Already standard in most office fit-outs.
3. Desk-mounted acoustic screens: 1.2 m minimum height, NRC 0.80+, between all adjacent workstations. Budget: $250-400 per screen.
4. Sound masking system: Plenum-mounted speakers on a 3 m x 3 m grid, calibrated to 42-44 dBA with ± 2 dBA spatial uniformity. Budget: $2-4/m².
5. HVAC noise control: NC-30 maximum at desk level. Requires lined ductwork, proper diffuser selection, and vibration isolation of mechanical equipment.
6. Partitions for enclosed rooms: STC 45+ from slab to slab. Full-height glazing should be laminated acoustic glass (minimum 10.76 mm laminated).

Generate Your WELL F74 Compliance Report

Calculating STI by hand is tedious and error-prone. AcousPlan automates the entire process: enter your room dimensions, select materials, specify workstation layout, and generate a WELL Feature 74 compliance report that shows RT60, background noise analysis, and STI calculations for every source-receiver pair in your open plan design.

The report includes pass/fail determination for all three Parts, identifies which workstations fail the STI threshold, and recommends the most cost-effective combination of absorption treatment and sound masking to achieve compliance.

Generate your WELL F74 compliance report with AcousPlan — model your room, run the simulation, and export a WELL-ready PDF in under 10 minutes.