A library in the early twentieth century was silent enough to hear a page turn three tables away. Every whispered conversation was audible to everyone within ten metres. Librarians enforced silence not because they enjoyed the authority but because without it, the room was acoustically unusable — every sound drew attention and disrupted concentration.
Today, most open-plan offices have the same problem but in reverse. They are spaces of relentless acoustic transparency. A phone call at one end of the floor reaches the ears of a colleague forty metres away. A spontaneous discussion between two people at adjacent desks becomes an involuntary participant experience for everyone within twelve metres. People resort to headphones, not because they want to listen to music, but because they need to introduce their own private acoustic shield.
Sound masking is the engineered solution to this problem. It works on a counterintuitive principle: to make a space feel acoustically quieter and more private, you deliberately add a carefully designed background noise. Not louder. Not more intrusive. Just enough to raise the floor of audibility so that distant conversations blend into ambience rather than demanding attention.
The Definition: What Sound Masking Actually Does
Sound masking is the introduction of a spectrally shaped background noise — typically delivered through loudspeakers mounted in or above the ceiling — at a level and frequency distribution chosen to render unintelligible speech that originates beyond a defined radius of distraction.
The key concept is the radius of distraction: the distance at which a person can understand the words being spoken at a nearby workstation. In a typical open-plan office with no sound masking and moderate absorption (RT60 approximately 0.5 seconds), this radius is approximately 10-15 metres. With a properly tuned masking system, it can be reduced to 4-6 metres — dramatically shrinking the area of acoustic interference around each worker.
Sound masking does not eliminate sound. It elevates the ambient floor — the minimum audible sound level in the space — so that conversational voices at medium distance are no longer above the threshold at which the brain involuntarily tries to decode them as speech.
The measure used to quantify this effect is the Articulation Index (AI) or its more modern successor, the Speech Transmission Index (STI), defined in IEC 60268-16:2020. An STI of 1.0 means perfect intelligibility. An STI below 0.45 means the speech is unintelligible — the listener cannot reliably decode words even at moderate attention. Sound masking works by reducing the signal-to-noise ratio between a distant voice and the background noise, pushing the STI in the receiving location below the intelligibility threshold.
ASTM E1130 (Standard Test Method for Objective Measurement of Speech Privacy in Open Offices Using Articulation Index) and ISO 3382-3:2012 (Acoustics — Measurement of Room Acoustic Parameters — Open Plan Offices) are the primary standards used to characterise and verify open-plan speech privacy performance, including the contribution of masking.
Why Shaped Noise Masks Better Than White Noise
Most people's first intuition about what kind of noise would mask speech best is "louder is better." But a masking signal that is too loud is simply annoying — it defeats the purpose by replacing one form of distraction with another. The goal is a masking signal that is barely noticeable but highly effective at the specific task of rendering speech unintelligible.
The answer lies in spectrum shaping. Speech energy is concentrated primarily in the frequency range from about 300 Hz to 3000 Hz, with the most intelligibility-critical energy between 500 Hz and 2000 Hz (where consonants — the sounds that distinguish words from each other — are concentrated). A masking signal that concentrates its energy in this same range can mask speech more efficiently than the same total sound level distributed across all frequencies.
Pink noise — a signal where each octave band contains equal energy, causing the spectrum to fall at 3 dB per octave — is a better masking signal than white noise (equal energy per Hz) because it more closely matches the spectral distribution of human speech and is less harsh-sounding. But most commercial masking systems use a purpose-tuned masking spectrum — pink noise with further adjustments to match the characteristics of the space's existing background noise, the height of the ceiling, and the directional properties of the loudspeaker array.
The goal is a masking signal that sounds like ordinary HVAC noise — something the brain classifies as "just the building talking" and stops attending to after a few minutes of acclimatisation. When a masking system is tuned correctly, most people in the space cannot consciously identify that it is present. They simply notice that the room feels less distracting and more private.
Where Sound Masking Is Used
Sound masking has applications wherever speech privacy is needed but full acoustic enclosure is impractical or undesirable.
Open-plan offices: The original and most common application. Modern work environments with low or no partitions, hard floors, and glass walls create near-complete acoustic transparency. Masking reduces the radius of distraction, improving concentration and reducing the social awkwardness of being involuntarily privy to a colleague's phone call.
Healthcare environments: Hospitals, clinics, and pharmacies are legally required in many jurisdictions to protect patient health information under regulations such as HIPAA in the United States. A reception desk or nursing station where staff discuss patient details within earshot of a waiting room is a compliance risk. Sound masking in these areas provides a practical privacy shield without requiring physical enclosure of every conversation.
Legal and financial services: Law offices, financial advisory firms, and HR departments require confidentiality as a professional matter. Conference rooms with inadequate STC ratings — particularly those adjacent to open office areas — can be supplemented with masking in adjacent zones to prevent sound leaking from the conference room from being fully intelligible in the adjacent space.
Government and defence secure facilities: Spaces where classified information is discussed may combine high STC partitions, acoustic isolation, and active masking systems as part of a layered speech privacy design.
Libraries and academic libraries: Counterintuitively, modern open-access libraries that contain both quiet study zones and collaborative learning areas benefit from masking in the study zones to reduce the intrusion of noise from collaborative spaces, without requiring physical separation of incompatible activities.
A Worked Example: Open-Plan Office Privacy Analysis
Consider an open-plan floor plate with the following characteristics:
- Room dimensions: 40 m x 20 m
- Ceiling height: 3 m
- Ceiling treatment: Acoustic mineral fibre tiles, NRC 0.75
- Floor: Hardwood with area rugs at workstations
- No partitions or screens
- Background noise (HVAC only): NC 30 (approximately 33 dBA)
At 8 metres from a talker, the speech level has decayed from approximately 65 dB (at 1 metre) to approximately 53-55 dB. The background noise level (NC 30) corresponds to approximately 33 dB(A). The signal-to-noise ratio at 8 metres is approximately +20 dB — more than enough for complete intelligibility. The STI at 8 metres is approximately 0.65 (Good). A colleague 8 metres away can understand every word of a phone call.
After adding sound masking tuned to NC 42 (approximately 44 dBA), the background rises to 44 dB(A). The signal-to-noise ratio at 8 metres is now approximately +9 dB. The STI drops to approximately 0.42 (Poor — speech is mostly unintelligible). The radius of distraction has shrunk from over 8 metres to approximately 4-5 metres.
The masking system has made the same speech that was fully intelligible at 8 metres essentially private, without changing any physical construction — no new walls, no new ceiling tiles, no reconfiguration of the floor plan.
Note: the masking level of NC 42 (44 dBA) is itself within the WELL v2 Feature 74 guideline for open-plan offices (which allows up to NC 40 for full credit but permits up to approximately NC 45 in zones where speech privacy outweighs quiet focus as a priority). If the masking level needed to achieve sufficient privacy would push the background above NC 45-50, that is a signal that the room needs additional absorption (to increase spatial decay rate) before masking can be effectively tuned.
Masking System Design: The Practical Elements
A sound masking installation involves more than simply playing a noise signal through ceiling speakers. Effective system design includes:
Loudspeaker placement: Most commercial masking systems use loudspeakers mounted above a suspended acoustic ceiling, pointing upward so that the ceiling tile acts as a distributed re-radiator. This produces a highly uniform sound field across the occupied floor area. Direct-field systems (downward-firing speakers) are also used but require careful tuning to avoid hotspots and dead zones.
Zoning: Large floor plates are divided into masking zones — typically 150-300 square metre areas served by a single set of amplifiers and a common equalization profile. Different zones may require different masking levels depending on their position relative to private offices, meeting rooms, or HVAC supply diffusers.
Equalization and tuning: After the system is installed, a qualified acoustic consultant or masking system technician measures the actual background level in the occupied zone at multiple positions and adjusts the system equalization and level until the target masking spectrum is achieved within approximately ±2 dB uniformity across the zone.
Time-of-day variation: Some advanced systems vary the masking level across the working day — louder during peak occupancy when natural background from people and activity partially contributes to masking, quieter in the early morning and late evening when the floor is nearly empty. This prevents the masking system from becoming consciously noticeable during quiet periods.
Integration with architectural acoustic treatment: Masking is most effective when combined with good room acoustic design — absorptive ceilings to increase the spatial decay rate of speech, screens and furniture to provide some physical obstruction, and partition designs that address the STC between enclosed offices and open areas. Masking alone cannot compensate for a room with excessive reverberation; a high RT60 means speech propagates farther before decaying, which means the masking system would need to operate at an uncomfortably high level to achieve privacy.
Common Mistakes When Specifying Sound Masking
Specifying the masking level without specifying the spectrum: A requirement stating "masking noise level shall be 45 dBA" is not adequately specific. Two masking systems with identical dBA levels but different spectral shapes can produce dramatically different speech privacy outcomes. The specification should reference a target masking spectrum (typically NC 40-45 or an octave-band limit table) and require octave-band verification after commissioning.
Installing masking without addressing reverberation first: In a room with RT60 above 0.8 seconds, speech propagates long distances before becoming unintelligible. The masking system must operate at much higher levels to overcome this effect. Reducing the RT60 first — by adding absorption — reduces the distance over which speech is intelligible, allowing the masking system to operate at comfortable levels.
Treating masking as a substitute for proper partition design: A masking system in the open area adjacent to a glass-walled conference room helps slightly, but it does not fix the fundamental problem of inadequate partition STC. Conversations inside the conference room are fully intelligible to anyone standing just outside if the glass has STC 30. The masking signal in the open area cannot cancel out that transmission. Masking supplements partitions; it does not replace them.
Not commissioning the system after installation: Many masking systems are installed and activated without octave-band measurement and professional tuning. The result is a system that may be too loud in some zones, too quiet in others, or spectrally mismatched to the room. Commissioning typically takes one to two hours per zone but is essential for achieving the specified privacy outcome.
How AcousPlan Helps You Design for Speech Privacy
AcousPlan's Speech Privacy Calculator models the interaction between room acoustics, masking systems, and partition design to give you a complete picture of privacy performance across a floor plate.
Within the calculator, you can:
- Model open-plan speech privacy using ISO 3382-3:2012 methods — calculate spatial decay rate, the D2,S parameter, and the radius of distraction before and after adding a masking signal at a specified NC level
- Assess the effect of different masking levels on STI at the receiver location, to find the minimum masking level needed to achieve the target privacy threshold without exceeding comfortable background noise limits
- Combine absorption, partitions, and masking in an integrated model, so you can determine the optimal investment split between acoustic ceiling treatment, screen height, and masking system level
- Check compliance against WELL v2 Feature 74 and ASTM E1130 privacy requirements
- Generate documentation showing predicted STI, spatial decay rate, and masking effectiveness for inclusion in design reports or WELL certification submittals
Ready to model your open-plan speech privacy? Try the Speech Privacy Calculator at /privacy — input your floor plate dimensions, ceiling treatment, and masking level, and get STI-based privacy assessment instantly.