What speech privacy actually means
Speech privacy is the degree to which a conversation in one part of a building cannot be understood by someone in another part. It is about whether words can be made out, not simply how loud they are.
Why a closed door is rarely enough
Picture the scene. You are in a meeting room with the door shut and the blinds drawn. The conversation is sensitive: a performance review, perhaps, or a discussion about a client in financial difficulty. You feel confident the room is doing its job. In all likelihood, it is not.
The reason is that privacy is about being understood, not about being quiet. A listener does not need to hear every word to follow a conversation, or to pick up something confidential. Partial intelligibility is enough. A standard interior door and a lightweight modern partition block far less sound than most people assume, and a quiet building gives that escaping speech very little to hide behind. So the words carry, and they remain just clear enough to follow.
Why modern offices are worse at this than they look
There is an uncomfortable irony here: when offices became quieter, speech became easier to overhear. As offices grew quieter — air-conditioning became near-silent, and the noisy tower PCs, keyboards, printers and fax machines of a generation ago disappeared — open-plan layouts replaced cellular rooms, and carpet tiles replaced hard floors, so the steady background sound that had always covered speech was quietly removed. Partitions were rarely upgraded to compensate. The result is offices which feel more acoustically comfortable but where conversations carry across an open floor, or straight through a partition, with surprising clarity.
This is also why the problem is so persistent. The brain treats speech differently from other sounds. A constant hum is quickly filed away as background, but speech is processed as meaning, so you cannot simply tune it out. That is not a failure of concentration. It is the listening system working exactly as it is built to. The good news is that the same fact points to the fix: if overheard speech can be made unintelligible, the brain stops locking onto it.
What it means for you
For most businesses, overheard conversations are a productivity, wellbeing and reputational issue: people are distracted, and sensitive matters are aired where they can be picked up. For organisations in regulated sectors, such as financial services, legal, healthcare, payroll and HR, there is a further dimension. Where client conversations, personal data or legally privileged discussions take place in shared spaces, adequate acoustic separation forms part of a proportionate approach to protecting information. We cover the picture regulator by regulator in our forthcoming compliance guides.
The three practical levers
Before the technical detail, the practical picture is reassuringly simple. There are three independent ways to improve speech privacy. Each one helps on its own, and used together they give the most predictable results.
- Sound masking — a carefully shaped electronic background sound, tuned to the frequencies of speech, that raises the ambient level so overheard talk becomes harder to follow. For privacy this is usually the simplest and cheapest first step: it makes overheard speech unintelligible without any structural work, and it is the fastest intervention in an existing building. We explain the technology in full in How Sound Masking Works.
- Partition performance. Upgrading, extending or sealing gaps in walls, glazing, doors and screens. This is the most permanent route, but also the most disruptive and expensive, and it only delivers if perfectly and fully implemented, which is rarely the case; it suits new builds and full refurbishments better than quick retrofits.
- Absorption. Adding acoustic panels, ceiling tiles and soft furnishings to soak up reflected sound, so speech fades faster as it travels. This improves comfort, but on its own it does not mask speech in the space next door.
You do not have to do all three, and they are not equally quick to deploy. For privacy, sound masking is usually the first step; partition improvements and absorption are then worth considering to make the space quieter overall — and because they leave the masking less to do, they make it work even better. Each lever helps on its own; combined they give the best and most predictable result, but where budget or the building allows only one, masking is the place to start.
The technical detail: how speech privacy is measured
Everything above can be specified, measured and verified. This second half sets out the metrics, standards and numbers that acoustic consultants actually design to.
The core metric: the Speech Transmission Index
The modern UK and European way to measure speech privacy is the Speech Transmission Index, STI — a measure of how much of what is said can actually be understood, expressed as a single number between 0 and 1. It takes account of background noise, speech level and reverberation, and is defined by the international standard IEC 60268-16 and used in ISO 9921. The principle is simple: a higher STI means speech is easier to follow, and a lower STI means more privacy. As a rough orientation, an STI around 0.6 or above represents clearly intelligible speech, while values toward 0.2 and below approach genuine confidentiality (Yadav and Cabrera, Routledge Handbook Ch.9).
Distraction distance and the open-plan standards
For open-plan and shared offices, two British Standards turn that index into practical design targets. BS EN ISO 3382-3:2022, the measurement standard for open-plan office acoustics, defines the most useful single figure for non-specialists: the distraction distance — the distance from a talker at which the STI falls to 0.5, beyond which speech is no longer meaningfully distracting. A short distraction distance indicates a space that controls speech well, and the design goal is under about 5 metres. The same standard defines the comfort distance (the distance at which a single talker’s speech level drops to 45 dB(A), the unit of loudness weighted to match human hearing), also desirably under 5 metres, and the spatial decay rate of speech, D2,S — how quickly speech level falls with each doubling of distance. These figures are drawn here via the Routledge Handbook chapter by Yadav and Cabrera, which cites ISO 3382-3:2022; they are the best available and should be validated against the standard itself.
BS ISO 22955:2021, the acoustic-quality standard for open offices, then sets the targets a space should meet. Per Yadav and Cabrera (Routledge Handbook Ch.9), citing that standard, the spatial decay of speech (D2,S) should be at least 7 dB in general offices and at least 8 dB in collaboration-focused spaces; reverberation time (how long a sound takes to fade away in a room, in seconds) should be roughly 0.5 to 0.8 seconds; and occupied sound levels should sit in the region of 48 to 55 dB(A) depending on use. BS 8233 provides the supporting UK guidance layer. Taken together, these standards describe what a genuinely private, or at least non-distracting, workplace looks like.
Why overheard speech is so disruptive: the mechanism
There is a specific, well-documented reason overheard speech is so hard to ignore. The Routledge Handbook of High-Performance Workplaces (2024), Chapter 9, ‘Acoustic privacy’, by Yadav and Cabrera (CC-BY-NC-ND 4.0), draws together two separate strands of evidence: acoustic measurements across 36 open-plan office floors, and an occupant survey of 426 participants. Both point to the same conclusion, consistent across more than fifty years of workplace studies: intelligible speech is the single greatest source of distraction and annoyance in open-plan offices.
The mechanism is specific. Speech fluctuates in intensity at roughly 4 Hz, about the rate of syllables. This syllabic rhythm, its fluctuation strength (a concept set out in Zwicker and Fastl, Psychoacoustics, 1999), triggers an involuntary attentional response. Unlike a steady hum, which the brain classifies as background, speech is processed as meaningful and decoded automatically. Hongisto (2005) models how this varying intelligibility translates into measurable losses in work performance. This is also why lowering STI works: when transmitted speech becomes unintelligible, the brain stops locking onto it.
A note on the US measure: SPP
Readers researching online, especially US sources, will encounter Speech Privacy Potential, SPP — an American manufacturer measure, SPP = STC + NC. It combines a partition’s Sound Transmission Class, STC (how much sound a partition blocks), with the Noise Criterion, NC (a rating of background noise level), on a 60 to 90 scale. It reflects a true intuition — that privacy depends both on what the partition blocks and on how much background noise covers the rest — but it is a US manufacturer convention rather than the standard the UK designs to. Where this article gives targets, they are expressed in STI and the ISO open-plan metrics above.
Who faces the greatest risk
For most businesses, overheard conversations represent a reputational and operational risk. For regulated businesses, they also represent a compliance exposure.
Organisations handling personal data under UK GDPR are required, under Article 32(1), to implement “appropriate technical and organisational measures to ensure a level of security appropriate to the risk”. The Information Commissioner’s Office confirms that physical security of premises is within scope. A controlled acoustic environment, one that keeps the distraction distance short and the STI low in sensitive areas, is one documented technical control that can support that obligation.
Financial services firms operating under FCA supervision must maintain effective systems and controls under the general systems-and-controls provisions of the FCA Handbook (SYSC 4.1.1R and 6.1.1R). Where client conversations take place in shared or semi-open spaces, acoustic controls can form part of a risk-assessed approach to protecting information.
Solicitors practising under the SRA Standards and Regulations have a duty under paragraph 6.3 of the Code of Conduct to “keep the affairs of current and former clients confidential”. A client call conducted in a meeting room with inadequate speech privacy is a potential weakness in that duty. HR teams, payroll functions and healthcare settings face similar considerations. Sound masking can form part of a broader compliance approach: it is one element of a wider framework, not a standalone solution.
Frequently Asked Questions
How do I know if my office has a speech privacy problem?
The simplest test is informal: stand in the space next to your meeting room with the door closed and listen. If you can follow the conversation, even intermittently, your speech privacy is inadequate. A more rigorous assessment measures the Speech Transmission Index (STI) and the distraction distance (the distance at which STI falls to 0.5) to BS EN ISO 3382-3:2022, which an acoustic consultant can carry out.
What is the Speech Transmission Index (STI)?
STI is a number from 0 to 1 that describes how intelligible transmitted speech is, accounting for background noise, level and reverberation. It is defined by IEC 60268-16 and used in ISO 9921. Lower values mean more privacy: values toward 0.2 and below approach confidentiality, while 0.6 and above means speech is clearly intelligible. It is the modern UK and European measure of speech privacy.
Is speech privacy a legal requirement?
There is no single regulation that states a required STI value. However, several frameworks, including UK GDPR Article 32, the FCA’s general systems-and-controls rules, and the SRA Code of Conduct, require organisations to take appropriate measures to protect confidential information, and a controlled acoustic environment can be part of a proportionate, risk-assessed approach. This article is for general information and is not legal or regulatory advice. Organisations should take their own professional advice.
Does closing the door solve the problem?
Partly, but rarely completely. A standard interior door and frame block only a modest amount of sound, and without adequate background noise to cover what passes through, the STI in the adjacent space can remain high enough for speech to be followed. A door that feels effective in a busy building can perform poorly once the building quietens — outside business hours, for instance.
What is the easiest way to improve speech privacy in an existing office?
For most existing offices, the lowest-disruption option is installing a sound masking system. Unlike partition upgrades, masking can typically be fitted in a single day without structural work. It raises the background level in a controlled way, shaped to the speech frequency range, so the gap between background noise and overheard speech narrows and the STI falls. We explain the technology in full in How Sound Masking Works.
Next steps
Sound Directions has worked with acoustic privacy in UK offices and regulated environments for over 25 years. If you would like to understand whether your current setup meets the speech privacy standard your sector requires, we would be glad to talk it through, without obligation.
Sources and further reading
Yadav, M., and Cabrera, D. (2024). ‘Acoustic privacy’. In Routledge Handbook of High-Performance Workplaces, Chapter 9 (pp. 102–115). Routledge. CC-BY-NC-ND 4.0. DOI: 10.1201/9781003328728-10. Open access.
Yadav, M., Cabrera, D., Love, J., Kim, J., Holmes, J., Caldwell, H., and de Dear, R. (2019). Reliability and repeatability of ISO 3382-3 metrics based on repeated acoustic measurements in open-plan offices. Applied Acoustics, 150, 138–146. Author preprint.
Hongisto, V. (2005). A model predicting the effect of speech of varying intelligibility on work performance. Indoor Air, 15(6), 458–468 (paywalled).
Zwicker, E., and Fastl, H. (1999). Psychoacoustics: Facts and Models (2nd ed.). Springer (paywalled).
Standards (official catalogue pages, paywalled): BS EN ISO 3382-3:2022; BS ISO 22955:2021; IEC 60268-16:2020; ISO 9921:2003; BS 8233.