Open Office Acoustics: Why Intelligible Speech Is the Problem and Absorption Alone Cannot Fix It

The complaint about open offices is usually phrased as noise, which points the solution in the wrong direction. Research on workplace distraction consistently finds that the disruptive stimulus is not sound level but speech intelligibility. A conversation you can understand at your desk captures attention involuntarily and degrades performance on any task involving language. The same conversation at the same level, but unintelligible, is substantially less disruptive.

This is why an open office can be quiet and still be unbearable to work in, and why adding absorption to a space that has a speech privacy problem often produces disappointing results. Absorption lowers the level of the intruding speech, but it also lowers the ambient level that was partially covering it, and the intelligibility can remain essentially unchanged.

This post covers the metric that actually describes open plan acoustic performance, the three-part method that addresses it, and why sound masking is the component most often omitted and most often decisive.

1. Intelligibility, Not Level, Is the Variable

Speech intelligibility depends on the relationship between the speech signal and the background noise, not on the absolute level of either. A voice 15 decibels above the background is fully intelligible whether that is 55 against 40 or 70 against 55. Drop the same voice to a few decibels above background and comprehension falls sharply.

Laboratory studies on cognitive performance in the presence of speech find that error rates on tasks involving reading, writing, and mental arithmetic rise substantially when intelligible speech is present, and that the effect tracks intelligibility rather than sound pressure level. Performance recovers as intelligibility falls, even when the speech remains audible.

The practical consequence is that the design objective in an open plan is not silence. It is to reduce intelligibility at neighbouring workstations to the point where overheard conversation registers as unremarkable background rather than as content the brain is compelled to process.

The metric

Open plan acoustic performance is quantified by the Articulation Index or, in current standards, the Speech Transmission Index and the distraction distance derived from it. Distraction distance is the distance from a speaker at which speech intelligibility drops below a defined threshold, conventionally an STI of 0.50.

This is a more useful design parameter than reverberation time, which describes an enclosed room's decay characteristics and does not translate well to a large open volume where the relevant question is how sound propagates horizontally between workstations rather than how it decays overall. ISO 3382-3 defines the open plan measurement method and the associated descriptors, including distraction distance and the spatial decay rate of speech.

Distraction distance

The distance from a talker at which speech intelligibility falls below an STI of 0.50. A well-performing open plan achieves under 5 metres. Poor spaces exceed 10 metres, which in a typical floor plate means most of the room can hear most conversations.

2. The Three-Part Method

Open plan acoustic design is conventionally described as absorb, block, cover. All three are required, and the common failure is treating the first as sufficient.

Absorb

Absorption removes sound energy, reducing reflections that would otherwise carry speech across the floor plate. The ceiling is the dominant surface because it is large, uninterrupted, and geometrically positioned to reflect sound from one workstation to another.

Ceiling absorption is specified by NRC or, more precisely for open plan, by the Articulation Class rating, which weights absorption performance in the frequency bands and incidence angles relevant to speech travelling between adjacent workstations. A ceiling tile with good NRC but poor AC can perform badly in open plan even though its overall absorption coefficient looks acceptable.

Where the ceiling is exposed structure, which is common in current workplace design, the absorption that a tile ceiling would have provided has to be replaced by suspended baffles, clouds, or applied absorptive treatment. Exposed structure with hard surfaces and no compensating absorption is one of the most reliable ways to produce an acoustically poor open plan.

Absorption elsewhere, on walls, on vertical screen surfaces, and in soft furnishings, contributes but cannot substitute for ceiling treatment because the geometry is less favourable.

Block

Blocking interrupts the direct sound path between workstations. The primary tool is the workstation screen, and its effectiveness depends on height, extent, and mass.

A screen must break the line of sight between a seated talker's mouth and a seated listener's ear to have meaningful effect on the direct path. Practically that requires a screen height of about 1.5 metres for seated-to-seated privacy. Screens at desk height, which are common because they preserve openness, block essentially nothing on the direct path.

The screen also has to have sufficient mass to attenuate transmission through it rather than simply reflecting, and absorptive facing to avoid reflecting sound back toward the talker's neighbours. A lightweight hard-faced screen can worsen conditions by adding reflecting surface.

And the screen only addresses the direct path. Sound reflecting off the ceiling passes over the screen entirely, which is why blocking without absorption performs poorly and why the two components are complementary rather than alternative.

Cover

Sound masking raises the background level with a continuous, spectrally shaped, unobtrusive sound distributed evenly across the space. Raising the background reduces the signal-to-noise ratio of intruding speech, which reduces intelligibility, which reduces distraction.

This is the component most often omitted, usually on the grounds that adding noise to solve a noise problem seems perverse. But it addresses the actual mechanism directly, and it is frequently the difference between a space that works and one that does not.

Why Masking Is Not Just Noise

Effective masking sound is engineered rather than arbitrary. The spectrum is shaped to be most effective in the frequency range that carries speech intelligibility while remaining unobtrusive, typically rolling off above about 4 kHz and shaped with a gentle downward slope from around 250 Hz. Level is typically between 42 and 48 dBA, tuned so that it is high enough to mask and low enough to go unnoticed. It must be spatially uniform, because variation is what makes masking perceptible; a system with audible level differences between areas draws attention to itself. And it should be commissioned by measurement across the floor plate rather than set by ear.

3. Why Absorption Alone Disappoints

A common sequence: an open office generates complaints, absorption is added to the ceiling and walls, and the complaints continue or worsen.

The reason is that absorption reduces both the intruding speech and the ambient background that was partially covering it. If both fall by similar amounts, the signal-to-noise ratio at the listener is unchanged and intelligibility is unchanged. The space is measurably quieter and subjectively no better, and in some cases worse, because the reduced background now allows conversations from further away to become intelligible.

This is the clearest demonstration that the design target is intelligibility rather than level. Absorption is necessary because it reduces the propagation of speech across the space and improves the spatial decay rate, but it needs to be paired with a controlled background level to produce the intended result.

4. Layout Decisions That Matter

Acoustic performance in open plan is influenced by planning decisions made well before any acoustic treatment is specified.

Separation distance.  Sound decays with distance, so workstation spacing directly affects the level of intruding speech. Densification, which is usually the reason an open plan exists, works against acoustic performance. This is a genuine trade-off rather than a problem to be engineered away, and it is better acknowledged at planning stage than discovered afterward.

Zoning by activity.  Placing collaborative areas, phone-heavy roles, and focused individual work in the same undifferentiated space guarantees conflict. Grouping by acoustic demand, with buffer zones such as circulation, storage, or plant between quiet and active areas, resolves much of it through planning rather than treatment.

Enclosed alternatives.  An open plan without adequate enclosed rooms for calls and meetings forces those activities into the open floor, which is the principal source of intruding speech. Providing sufficient focus rooms, phone booths, and small meeting spaces removes the source rather than treating its effects. Under-providing them is a common and consequential planning error.

Orientation and reflecting surfaces.  Workstations facing each other across a low divider propagate speech directly. Hard reflective surfaces near workstations, glass partitions, whiteboards, exposed columns, redirect sound in ways that can defeat screen placement. Acoustic modelling during layout development catches these.

5. Speech Privacy in Enclosed Rooms Is a Different Problem

Enclosed rooms within an open plan office fail acoustically in predictable ways, and the failure is usually not the wall itself.

A partition rated STC 45 delivers that performance only if the assembly is continuous and sealed. Where the partition stops at a suspended ceiling and the plenum above is continuous to the adjacent room, sound passes over the top and the effective performance drops to whatever the ceiling and plenum path allows, frequently under STC 30 regardless of the wall rating.

The other common paths are unsealed penetrations for conduit and piping, back-to-back electrical boxes in the same stud cavity, continuous ductwork connecting two rooms without a lined transfer path or crosstalk attenuator, and door undercuts. A partition carried to structure with all penetrations sealed and a properly gasketed door will perform close to its rating; the same partition with any one of those paths open will not.

For rooms requiring confidential speech privacy, the design target is stated in terms of a composite metric such as NIC or, where verified in the field, ASTC, measured on the completed room rather than assumed from the partition specification.

Conclusion

Open plan acoustic design fails when it is framed as a noise problem. The disruptive stimulus is intelligible speech, the governing metric is distraction distance derived from speech intelligibility, and the design objective is to make overheard conversation unintelligible rather than inaudible.

That requires all three components. Absorption, concentrated at the ceiling and specified by an open plan appropriate metric, reduces propagation across the floor plate. Screens of sufficient height and mass break the direct path. And a properly engineered masking system provides the controlled background that makes the remaining intruding speech unintelligible. Absorption alone lowers both signal and background and frequently changes nothing.

Underneath the treatment, planning decisions about density, activity zoning, and provision of enclosed alternatives determine how much the acoustic design has to accomplish. Getting those right reduces the problem; getting them wrong sets a limit on what any amount of treatment can achieve.