8 The indoor and the personal
The machine hum did not stop at the wall; it settled into rooms. Indoors, environmental noise changes what it costs. On the street it is counted in premature deaths and broken sleep. In the rooms where people learn and work it is counted in something quieter and far harder to see — a sentence half-heard, a paragraph read twice, a train of thought broken and not recovered. Here noise is rarely loud. It is the ventilation drone of an open-plan office, the hard echo of an under-treated classroom, the traffic bleeding through a single-glazed window, the private wash of a pair of headphones. What it degrades is intelligibility and attention rather than comfort, and the children still learning to read carry that cost least evenly. The temptation is to let noise stand in for distraction in general. The discipline here is to keep it acoustic, as a measurable signal competing with a measurable background, and to follow what that competition does to understanding.
8.1 Intelligibility, not loudness
What matters indoors is not the absolute loudness of the background but the margin between it and the speech a listener is trying to follow: the speech-to-noise ratio. Alongside it stands the time the room takes to fall quiet after a sound, its reverberation time. Both are ordinary acoustic measurements; neither is a metaphor. A library at forty decibels and a canteen at forty decibels are identical on a sound-level reading and worlds apart in what can be understood in them, because understanding depends on the relation between voice and background, not on the background alone.
The two quantities degrade comprehension by different routes. A high background masks the quieter parts of speech directly, burying the consonants that carry most of its meaning while the louder vowels, which carry less, survive. A long reverberation time does something subtler: it smears each sound forward in time, so that the tail of one syllable overlaps the onset of the next and the boundaries between words dissolve. A room can therefore be perfectly quiet and still hard to follow, if it rings; the two faults are independent, and a space has to control both. Acousticians fold them into a single figure, the Speech Transmission Index, which runs from zero to one and rises as the combination of a favourable speech-to-noise ratio and a short reverberation time preserves the modulations of speech on their way from mouth to ear. The index is banded for interpretation: a value above roughly 0.75 counts as excellent and a value below about 0.45 counts as poor, with the range between them marking the distance between a room a listener follows without effort and one in which every sentence has to be partly reconstructed (International Electrotechnical Commission, 2020).
The recommended values for a classroom are unambiguous. The international standard sets a background at or below 35 dB(A) in the unoccupied room, a reverberation time at or below 0.6 seconds, and a speech-to-noise ratio of at least fifteen decibels at the pupil’s ear (Acoustical Society of America, 2010). That combination corresponds to a Speech Transmission Index of about 0.6, the floor below which intelligibility is judged inadequate for learning. The domestic legal objective sits above it. Spain’s noise regulation fixes the interior target for a classroom at 40 dB, five decibels above the level the acoustic evidence recommends (BOE, 2007). The gap is small in decibels and large in consequence, and the pattern is one already met more than once — the number written into law sitting a comfortable margin away from the number the evidence would set. The same index recurs in every room that follows. What changes, from the classroom to the office, is not the physics but the goal.
8.2 The classroom
A child does not listen the way an adult does. An adult who misses a word rebuilds it from the surrounding sentence, from grammar, from expectation; a seven-year-old still assembling those resources has less to fall back on, and so needs a cleaner signal to reach the same understanding. This is why the margin recommended for a classroom is wider than an adult would require — a speech-to-noise ratio of at least fifteen decibels (Acoustical Society of America, 2010). Some children need it wider still. A child learning in a second language, or with a hearing loss, a developmental disorder, or the fluctuating middle-ear congestion common in early childhood, has even less spare capacity to reconstruct a degraded signal; the acoustic conditions that merely inconvenience a confident native speaker can shut such a child out of the lesson altogether. A standard written for the average pupil is, for them, already a compromise (Acoustical Society of America, 2010). The cost of failing to provide it is a chronic tax on the effort of listening, levied over the years in which reading and attention are being built.
The mechanism is one of effort. Following speech in noise is not automatic; it draws on working memory and sustained attention to hold the fragments of a sentence while the missing pieces are inferred. A favourable acoustic environment makes that reconstruction cheap, and a poor one makes it expensive, levying a continuous charge on the same cognitive resources a lesson is meant to build. For an adult the charge is usually affordable. For a child it competes directly with the work of learning to read, and because the exposure is chronic the small daily costs accumulate into a measurable lag — the same classroom, the same road outside, day after school day. The pathway does not run through attention alone. The autonomic and stress responses described in Chapter 3 operate here too, so that a noisy classroom taxes a child bodily as well as cognitively, throughout the years in which the capacities under strain are still forming.
The evidence that the tax is real has accumulated steadily. The systematic review prepared for the World Health Organization’s environmental noise guidelines examined thirty-four studies of noise and cognition, every one of them conducted on children, and found evidence of moderate quality that chronic exposure to transport noise is associated with poorer reading comprehension and impaired memory (Clark & Paunović, 2018). Applied across the continent, those relations imply a burden that is not small. European environmental agency estimates attribute to transport noise more than half a million cases of reading-comprehension impairment among European children, alongside tens of thousands of cases of behavioural difficulty (European Environment Agency, 2025).
Two studies anchor the association in Spain, and they complement each other. The older, foundational one is the RANCH project, which measured aircraft and road-traffic noise at schools around three European airports (Heathrow, Schiphol and Madrid-Barajas) and related exposure to children’s reading comprehension, establishing an exposure-effect relation in which comprehension declined steadily as noise rose, consistently across the three countries (Clark et al., 2006). The more recent is a Barcelona cohort that followed 2,680 primary-school children, aged seven to ten, across thirty-eight schools over a year, tracking the development of their working memory and attention against the road-traffic noise measured at their schools. Children at the noisier schools progressed more slowly on every cognitive test; the association held for noise at school but not for noise measured at the children’s homes, which points to the classroom, where the sustained work of learning is done, as the place where the exposure bites (Foraster et al., 2022). One detail of that study reaches past the classroom. What tracked most closely with slower progress was not the average noise level but its fluctuation — the peaks (Foraster et al., 2022).
The cost is not distributed evenly, and its unevenness is not accidental. Land beside a motorway, a rail corridor or a flight path is cheaper, and the schools and housing built on it more often serve families with the least room to choose otherwise, so that the acoustic burden falls, with grim regularity, on the children for whom an additional handicap is hardest to carry. The pattern is not universal, and a case that runs the other way is set out in Chapter 9. But for the transport noise that fills a child’s school day, exposure and disadvantage tend to move together.
Granada has no study of its own. What it has is an exposure map. The city’s strategic mapping puts fewer than about 26,000 residents (under eleven per cent of a resident population of roughly 234,000) above 65 dB in the day–evening–night average (El Independiente de Granada, 2023; INE, 2025). That ceiling, though, is set against the permissive national quality objective, which sits above the levels the World Health Organization treats as health-relevant. Measured against the WHO values, road traffic alone leaves close to a third of the city above 50 dB at night, and more than half above 55 dB by day — around 78,000 residents affected at night on that reckoning (El Independiente de Granada, 2023).
Its schools sit within that same acoustic geography. For those on the most exposed streets, the exposure-effect relations found in Barcelona and around Madrid would predict, on the same mechanism, a comparable cost to working memory and attention among their pupils — an expectation, not a measurement, and one not yet tested locally.
8.3 The office and the headphone
The adult indoor world turns the classroom problem on its head. In a school the aim is to make one voice, the teacher’s, as intelligible as possible to everyone in the room. In an open-plan office the aim is the reverse: to make the many voices at neighbouring desks as unintelligible as possible to everyone not party to them. The same physical quantity governs both, but the target is inverted, and the reason lies in a well-documented quirk of attention. It is not loud speech that most degrades cognitive work, but intelligible speech. A conversation that can be made out — wanted or not — recruits the language faculties involuntarily and competes for the very resources a written task requires; the same conversation reduced to an unintelligible murmur largely stops interfering. The disturbance tracks intelligibility, not volume.
The room-acoustic standard for open-plan offices is built directly on this finding. It defines a distraction distance: the distance from a speaker at which the Speech Transmission Index of their voice falls below 0.50, the point beyond which their speech is no longer intelligible enough to seriously disrupt a neighbour (International Organization for Standardization, 2022). The same standard also describes how quickly speech should fade across a floor, measuring the rate at which its level falls with each doubling of distance from the talker. A high rate of decay — won through absorptive ceilings, screens and soft floors — is treated as the mark of a well-designed open plan. The aim is counter-intuitive. A quiet office would leave every telephone call legible from thirty desks away; the standard asks instead for a background raised just enough, and a decay made fast enough, that the words at any distant desk arrive already blurred.
How much is at stake is itself quantified. A model widely used to set the criterion relates the decrement in cognitive performance to the intelligibility of the intruding speech, and finds the relation steepest across the mid-range of the index — performance falls away rapidly as intelligibility climbs from about 0.25 to about 0.60, then flattens. The penalty is bounded but not trivial: at the worst intelligibilities the modelled loss on a demanding task settles at around seven per cent, a tax paid quietly and continuously by everyone on the floor (Hongisto, 2005). It follows that the two indoor problems demand opposite remedies from the same firm. The money spent on a training room or a workplace crèche goes to raise the index, pulling a single voice clear of the background; the money spent on the open floor next door goes to lower it, pushing every voice down into indistinctness. Absorption, screening and deliberately added masking sound all serve the second end — indistinctness at the next desk.
The headphone is the individual’s answer to both problems, and it carries its own. Worn at a desk, it is a refuge — a private soundscape that masks the office and restores some of the control the open plan took away. Worn on the street or the train, it is an exposure. The volumes people choose to overcome ambient noise are frequently high enough to matter: the World Health Organization estimates that around 1.1 billion young people are at risk of hearing loss from unsafe listening, roughly half of them through personal audio devices (WHO, 2019). The boundary of safe listening is set at 80 decibels for adults, and at 75 for children, over a working week’s exposure (World Health Organization & International Telecommunication Union, 2019). The dynamic is self-reinforcing, for the louder the surroundings, the higher the volume a listener sets to ride above them, so that the noisiest environments draw the largest self-administered doses — the very environments headphones are used to escape. The safe dose also falls steeply as the volume climbs — about forty hours a week at 80 decibels, roughly four hours at 90, some twenty minutes at 100. A level chosen to override a loud train therefore buys minutes rather than hours before it begins to wear on hearing. That is why the safe-listening standard now adopted for personal devices builds in dose tracking and volume limits instead of trusting a listener’s own sense of what is loud (World Health Organization & International Telecommunication Union, 2019). A systematic review pooling data from across twenty countries put the global number of adolescents and young adults potentially at risk, from personal devices and loud venues together, at somewhere between 0.67 and 1.35 billion (Dillard et al., 2022). The instrument most people reach for to take command of their own soundscape is thus also, for a great many of them, a self-administered dose — the private end of a public problem, and the point at which the exposure becomes something a person does to themselves rather than something done to them.
8.4 The classroom simulator
The classroom is where the trade-off matters most, and it is the case modelled below. The Speech Transmission Index responds there to three settings at once: the background noise level, the reverberation time, and the level at which the teacher’s voice reaches a nearby child. A quiet room proves insufficient on its own, because a long reverberation time degrades intelligibility even when the background is low. The standard’s three requirements have to be met together, not traded one against another. The point where the index crosses 0.6, the classroom floor, is the point at which a lesson stops being fully followable for the children who most need it to be.
Source · speech-to-noise ratio and reverberation combined into a Speech Transmission Index after the simplified modulation-transfer method of International Electrotechnical Commission (2020); classroom target values (background ≤ 35 dB(A), reverberation ≤ 0.6 s, speech-to-noise ≥ 15 dB, STI ≥ 0.60) after Acoustical Society of America (2010). Accessed 30 July 2026.
8.5 The cost that does not show up on a metre
A strategic noise map records the level at a façade. It does not record whether the child behind that façade finished the paragraph, or whether the clerk two floors up read the clause twice, or how much of the effort a room demanded was spent on hearing rather than on thinking. The indoor cost of noise is real, it is measurable in principle, and it is almost entirely absent from the monitoring that governs environmental noise, which stops at the outer wall and reports an annual average of what it finds there. The quantity that does the damage indoors is the moment-to-moment margin between a voice and its background — the quantity outdoor mapping is not built to see.
The Barcelona finding sharpens the point. It was the sharpest fluctuation, not the highest average, that tracked the children who fell furthest behind (Foraster et al., 2022). An indicator built by averaging sound energy over a day, a year, a lifetime smooths those peaks into a single background figure and reports the room as quieter than the experience of being in it. The same operation, applied to a different room, let a steady tone disappear from a compliance reading in Chapter 7. Whether the source is a data centre or a passing lorry, the averaging that makes a metric governable also makes it partly blind to the feature that carries the cost.
That cost does not stay acoustic for long. A reading age not reached, an attention span not built, a working memory a year behind where it might have been — these convert, over a lifetime, into measurable disadvantage in earnings, in health, in the opportunities a person can take. And they fall, more often than not, on the same people twice over: the household beside the busy road absorbs the outdoor burden of disturbed sleep and raised cardiovascular risk, and sends its children to the school where the indoor burden is heaviest, the two exposures compounding instead of cancelling. What that compounded cost is worth, who ends up paying it, and whether the arithmetic could justify the price of quieter schools and quieter offices, are questions of value rather than acoustics. They are where Part III begins.