1 What counts as noise
Ask what noise is and the intuitive answer arrives before the question has finished: noise is unwanted sound. The definition is serviceable enough, and it has organised environmental regulation for the better part of a century. But notice what it smuggles in. Unwanted is not a property of the air. It is a judgement made by a listener, in a place, at a time, and the same physical signal can cross the line between wanted and unwanted depending on who is listening and why. The train that is infrastructure to the commuter on the platform is an assault to the shift-worker trying to sleep beside the track. Nothing about the sound has changed; the person and the context have. The definition that seems to locate the problem in the sound has, on inspection, located it in the relation between the sound and the hearer.
That relation is what the last two decades have turned into an object of study. For most of its history, noise control treated sound as a quantity emitted by a source and received at a point — something to be metred, capped and abated, with the listener present only as a threshold to be protected. The soundscape turn did something more radical than add nuance. It moved the object of study from the sound to the experience of the sound, and in doing so gave noise a vocabulary precise enough to be measured, mapped, and eventually designed. That move is the foundation the rest of this work is built on, and it is also, quietly, the first appearance of a theme that will run underneath everything that follows: what we choose to measure decides what we are able to see.
1.1 Unwanted sound, and why the definition fails
The classical account has an honourable pedigree and a real use. Defining noise as unwanted sound draws a bright line between acoustics and annoyance, leaving the physicist to measure the pressure variation and the regulator to decide how much of it a person should have to tolerate. It underwrites limit values, working-time exposure rules and the whole apparatus of noise abatement. Within its domain it works.
The definition fails not by being wrong but by being incomplete, in a way that matters more the closer one looks. Three things escape it.
First, the same sound is wanted and unwanted at once, depending on the listener. A church bell, a call to prayer, a children’s playground, a music venue — each is cherished by some of the people within earshot and resented by others, and no measurement at the property line can adjudicate between them because the disagreement is not about decibels. It is about meaning.
Second, the definition cannot explain why sounds of equal level are not equally unwanted. A steady mechanical hum and the varied sound of a street market can register the same figure on a meter and be experienced as opposites — one draining, one enlivening. If noise were simply unwanted sound, and unwantedness tracked level, this could not happen. It happens constantly.
Third, and most consequentially for this work, the definition takes the source as given and asks only how loud it is. Nothing in it can ask what a place should sound like — whether a sound belongs, whether its absence would be an improvement or a loss. It can subtract; it cannot compose. A framework that can only ever say less has already given up the more interesting question.
The same element is missing from all three failures — the listener, the context, the meaning. To restore them is not to abandon acoustics but to recognise that noise lives in the meeting of a physical environment and a perceiving person. The soundscape framework is the attempt to make that meeting an object of rigorous study.
1.2 The soundscape turn
The word soundscape carries an older, evocative history — the acoustic-ecology tradition that taught a generation to hear the environment as a composition rather than a nuisance to be suppressed. But an evocative idea is not yet a measurable one, and for decades the term stayed idiosyncratic; everyone meant something slightly different by it, which is another way of saying that no one could build on anyone else’s work.
The decisive move was to fix the definition. In 2014 the International Organization for Standardization published the first part of a three-part series that did exactly this. ISO 12913-1 defines a soundscape as the acoustic environment as perceived or experienced and understood by a person or people, in context (International Organization for Standardization, 2014). The phrasing is doing careful work, separating two things that ordinary language runs together. The acoustic environment is the physical phenomenon, the sound field, measurable in the usual way. The soundscape, by contrast, is a perceptual construct that exists only through a listener’s engagement with that environment. The acoustic environment is out there whether or not anyone hears it. The soundscape is not; it comes into being in the hearing.
That distinction looks like a philosopher’s nicety and is in fact the hinge of the whole field. Once perception is written into the definition, context stops being an inconvenience to be controlled away and becomes part of the object. The standard is explicit that a soundscape assessment is through human perception of the acoustic environment, and it lists the factors that shape that perception: the listener’s attitude to the source and to whoever is producing the sound, their expectations and cultural background, their reason for being in the place, their sense of whether they can control their exposure, and the other senses feeding in alongside the ears. None of these appears on a sound-level meter. All of them determine whether a given acoustic environment is experienced as pleasant or as intolerable.
The framework, in other words, does not soften the physics; it embeds the physics in the situation where sound is actually encountered. This is what allows later chapters to speak of designing a soundscape rather than only suppressing sound — a possibility taken up in full in the constructive turn of the final chapter. But it also sets up the more critical thread. If soundscape is perception-dependent, then any instrument that claims to measure it is making a claim about perception, and the choices built into that instrument — which frequencies it weights, which it discards — become claims about what is worth perceiving. The next chapter is about one such instrument and the choices hidden inside it.
1.3 The eight attributes
A definition that stays qualitative changes nothing in practice. The real achievement of the soundscape programme was to make perception quantitative without flattening it — to find a small number of dimensions that capture how people actually experience acoustic environments, and a defined set of scales for measuring them.
The empirical basis came from a study by Axelsson, Nilsson and Berglund, who had a hundred listeners rate fifty binaural recordings of urban outdoor environments on more than a hundred attribute scales, then reduced the results to their underlying structure (Axelsson et al., 2010). Three dimensions emerged, but two did almost all the work: a pleasantness dimension, accounting for roughly half the variance, and an eventfulness dimension accounting for close to a fifth. (A third, familiarity, explained little and is usually set aside.) The finding is more than a statistical convenience. It says that when people judge a soundscape, they are answering two largely independent questions — is this pleasant or unpleasant? and is much happening here or little? — and that most of the perceptual richness of everyday acoustic experience lives in the plane those two axes define.
This is the point at which attribution has to be exact, because most summaries blur it. The two-dimensional pleasant–eventful structure is Axelsson and colleagues’ finding. The now-standard eight perceptual attributes are the operationalisation of that structure in the second part of the ISO series. In its Method A, ISO/TS 12913-2 sets out eight descriptors — pleasant, vibrant, eventful, chaotic, annoying, monotonous, uneventful and calm — rated on bipolar five-point scales, as the recommended way to collect soundscape data in the field (International Organization for Standardization, 2018). The eight are not eight independent qualities. They are eight equally spaced positions around a circle, the circumplex, whose two underlying axes are pleasantness and eventfulness. The third part of the series, ISO/TS 12913-3, then specifies the geometry that makes this usable. Neighbouring attributes sit forty-five degrees apart and opposites a hundred and eighty, and a set of ratings can be transformed into a single coordinate in the pleasant–eventful plane (International Organization for Standardization, 2019).
Reading around the circle shows why the eight attributes are the right eight rather than an arbitrary list, and the structure turns out to be intuitive. A vibrant soundscape is one that is both pleasant and eventful — a lively square that people enjoy. Rotate towards pure eventfulness and then into unpleasantness and you reach the chaotic: eventful but not wanted, the noisy street corner. Continue to the monotonous, which is both unpleasant and uneventful — the draining drone with nothing in it. Round again to the calm, both pleasant and uneventful, the quiet park. Each of the eight names a compass point, and any real soundscape can be placed by where its ratings fall in the plane (Aletta et al., 2016). The standard preserves one nuance deliberately. In the environmental-noise tradition the pole opposite pleasant is labelled annoying rather than the more neutral unpleasant, keeping a bridge to the older language of annoyance research even as the framework moves beyond it.
Source · perceptual axes and eight attributes after Axelsson et al. (2010) and ISO/TS 12913-2:2018 (International Organization for Standardization, 2018); circumplex geometry per ISO/TS 12913-3:2019 (International Organization for Standardization, 2019). Accessed 2026-07-11.
The simulator above makes one claim tangible. The same recorded sound occupies a definite position in a perceptual space, and that position — not its raw level — is what determines whether it is experienced as noise. Two sounds of identical loudness can sit at opposite ends of the pleasantness axis. That is the soundscape framework’s central empirical fact, and the one the decibel, on its own, cannot represent.
1.4 What a definition decides
The argument here is easily mistaken for a plea for subtlety — for remembering that noise is complicated and perception matters. It is meant as something more pointed. To define noise relationally, as the soundscape framework does, is already to have made a decision with consequences that run through the rest of this work, because a definition is not a neutral starting point. It sets the terms on which everything downstream will be measured, mapped and governed.
Consider what follows from the two ways of defining the object. If noise is unwanted sound at a source, the natural instruments are a level meter and a limit value, the natural question is how loud, and the natural response is abatement. If noise is a soundscape — a perceived environment in context — the natural instruments include the perceptual scales just described, the natural question becomes how is this place experienced, and the natural response opens onto design as well as reduction. These are not two descriptions of one practice. They are two different practices, and they make different things visible.
Much of what follows lives in the gap between them. The metric that governs European noise policy sits firmly on the first side of this divide — a single weighted number, source-oriented, blind by construction to much of what the soundscape framework was built to capture. That is not an oversight to be tidied up. It is the shape of a choice about what counts, and the sources changing our acoustic world fastest are precisely the ones that choice cannot see. The soundscape turn matters here not because it is more humane but because it is more honest about a fact the older definition obscured: that before we measure noise, we have already decided what noise is, and that decision does work no measurement can undo.