ISO/TS 12913-3:2025

Technical
Specification
ISO/TS 12913-3
Second edition
Acoustics — Soundscape —
2025-05
Part 3:
Data analysis
Acoustique — Paysage sonore —
Partie 3: Analyse de données
Reference number
© ISO 2025
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on
the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address below
or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: +41 22 749 01 11
Email: copyright@iso.org
Website: www.iso.org
Published in Switzerland
ii
Contents Page
Foreword .iv
Introduction .v
1 Scope . 1
2 Normative references . 1
3 Terms and definitions . 1
4 General . 1
5 Analysis of quantitative data . 2
6 Analysis of qualitative data . 2
7 Analysis of binaural data . 2
8 Triangulation . 3
Annex A (informative) Analysis of data related to Method A . 4
Annex B (informative) Analysis of data related to Method B .12
Annex C (informative) Analysis of data related to Method C . 14
Annex D (informative) Analysis of binaural data .16
Annex E (informative) Triangulation .20
Bibliography .21

iii
Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out through
ISO technical committees. Each member body interested in a subject for which a technical committee
has been established has the right to be represented on that committee. International organizations,
governmental and non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely
with the International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are described
in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the different types
of ISO document should be noted. This document was drafted in accordance with the editorial rules of the
ISO/IEC Directives, Part 2 (see www.iso.org/directives).
ISO draws attention to the possibility that the implementation of this document may involve the use of (a)
patent(s). ISO takes no position concerning the evidence, validity or applicability of any claimed patent
rights in respect thereof. As of the date of publication of this document, ISO had not received notice of (a)
patent(s) which may be required to implement this document. However, implementers are cautioned that
this may not represent the latest information, which may be obtained from the patent database available at
www.iso.org/patents. ISO shall not be held responsible for identifying any or all such patent rights.
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and expressions
related to conformity assessment, as well as information about ISO's adherence to the World Trade
Organization (WTO) principles in the Technical Barriers to Trade (TBT), see www.iso.org/iso/foreword.html.
This document was prepared by Technical Committee ISO/TC 43, Acoustics, Subcommittee SC 1, Noise.
This second edition cancels and replaces the first edition (ISO/TS 12913-3:2019), which has been technically
revised.
The main changes are as follows:
— editorial changes throughout the document;
— revision of Annex A;
— suppression of Annex F.
A list of all parts in the ISO 12913 series can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www.iso.org/members.html.

iv
Introduction
The ISO 12913 series on soundscape was developed in order to enable a broad international consensus
and to provide a foundation for communication across disciplines and professions with an interest in
soundscape. ISO 12913-1 provides the definition of and a conceptual framework for the term ‘soundscape’.
ISO/TS 12913-2 provides requirements and supporting information on data collection and reporting for
soundscape studies, investigations and applications. This document provides guidance on how to analyse
data collected in agreement with ISO/TS 12913-2.

v
Technical Specification ISO/TS 12913-3:2025(en)
Acoustics — Soundscape —
Part 3:
Data analysis
1 Scope
This document provides requirements and supporting information on analysis of data collected in situ
through methods as specified in ISO/TS 12913-2.
2 Normative references
The following documents are referred to in the text in such a way that some or all of their content constitutes
requirements of this document. For dated references, only the edition cited applies. For undated references,
the latest edition of the referenced document (including any amendments) applies.
ISO 12913-1, Acoustics — Soundscape — Part 1: Definition and conceptual framework
ISO/TS 12913-2:2018, Acoustics — Soundscape — Part 2: Data collection and reporting requirements
3 Terms and definitions
For the purposes of this document, the terms and definitions given in ISO 12913-1 and ISO/TS 12913-2 and
the following apply.
ISO and IEC maintain terminology databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at https:// www .iso .org/ obp
— IEC Electropedia: available at https:// www .electropedia .org/
3.1
confounder
factor influencing the collected responses that is not controlled or systematically considered
EXAMPLE Sequential effects, certain scaling effects such as the range effects, or demand characteristics.
4 General
As mentioned in the Introduction of ISO/TS 12913-2:2018, “The concept of soundscape was adopted to provide
a holistic approach to the acoustic environment, beyond noise, and its effect on the quality of life. Soundscape
investigations intend to assess all sounds perceived in an environment in all its complexity. To do this, soundscape
studies use a variety of data collection methods related to human perception, the acoustic environment and
the context. Importantly, the study of soundscape relies primarily upon human perception, and only then turns
to physical measurement.” Data collection is based on this focus and requires a respective analysis (see
References [1][2]).
For the analysis of qualitative and quantitative data through methods specified in ISO/TS 12913-2, methods
and tools shall be applied as provided in this document. Given the diversity of the data collected (qualitative
and quantitative), corresponding analysis methods could take precedence depending on the needs of the
project or the research question, and should be integrated for a holistic understanding of the soundscape.

In general, descriptive statistics are used to describe and summarize the collected perceptual data, such as
measures of central tendency, measures of dispersion (see Reference [3]).
For quantitative measures of dependence, inferential statistics using parametric and non-parametric
tests shall be applied depending on the respective data. Fulfilment of model assumptions (e.g. normality
distribution) shall be carefully assessed, especially in the case of small numbers of participants. If needed,
appropriate remedial measures shall be applied. However, as soundscape investigations are intended to be
“holistic in covering all auditory sensations as well as all other context variables such as visual stimuli and
personal expectations” [ISO/TS 12913-2], the use of statistical analysis methods (e.g. statistical hypothesis
testing) is recommended, but may be less important in case of qualitative or explorative methods. For
qualitative data, a variety of approaches are available to systematically analyse qualitative data using some
kind of step by step coding principles to generalize the observations.
Because of factors that could influence results, a thorough discussion of potential confounders (i.e. bias
effects) shall complete the general data analysis. Confounders are, for example, the sequential effect (a
previous site influences the assessments of the following site) (see Reference [4]) certain scaling effects, like
the range effect (tendency to use full range of a scale independent from stimuli set) (see Reference [5]), or
demand characteristics (cues, like the instruction text or the behaviour of the person leading the soundwalk,
that signal the research goal and influence assessments) (see Reference [6]).
NOTE Based on the collected data, it is possible to study the classification of sites. For the study of classification
of sites, different statistical clustering methods are available, which allow for identifying relevant variables for
clustering and determining the similarity or dissimilarity of sites, based on both acoustical data (see Reference [7])
and perceptual data (see Reference [8]).
5 Analysis of quantitative data
The quantitative data obtained by means of questionnaires in soundscape investigations shall be analysed
depending on the respective level of measurement (i.e. nominal, ordinal, interval, and ratio). Any correlation
analysis shall be chosen in accordance with the level of measurement of the questionnaire data. Inferential
statistical tests regarding the level of significance of differences in evaluation between sites and/or
correlations shall be carried out and probability values reported. Any chosen method (e.g. measure of central
tendency, measure of dispersion, correlation analysis, and statistical hypothesis testing method) shall be
reported. For more information, see Annex A (Method A) and Annex B (Method B).
6 Analysis of qualitative data
Data from qualitative interviews shall be transcribed for reporting and further analysis. The style of
transcription, whether clean read, verbatim or strict verbatim transcription, depends on the object of the
investigation. For the analysis via the Grounded Theory, the clean read transcription style is sufficient.
Violations of common rules for conducting interviews (ethical rules, being suggestive, being prejudiced)
shall be reported, and the related data excluded from further analysis.
Qualitative data shall be analysed by scientifically proven systematic text analysis methods, such as the
Grounded Theory (see Reference [9]), Qualitative Content Analysis (see Reference [10]) or Social Network
Analysis as part of mixed-methods design (see Reference [11]). The process of analysis shall follow these
methods and be described. For more information, see Annex C (Method C).
In addition to established text analysis methods, other methods to gather and analyse qualitative data
(such as behavioral mapping, observational analyses, analysis of social interaction, walking patterns; see
examples in References [12][13][14][15]) are available and, if determined appropriate in certain cases, shall
be applied.
7 Analysis of binaural data
The binaural recordings are the basis for characterizing the acoustic environment at the receiver as the
sound from all sound sources modified by the environment [ISO 12913-1]. The measurements and their
psychoacoustic analyses enable the determination of the (basic) auditory sensations evoked by the sound.

Any binaural recording shall be equalized for the analysis as specified in ISO/TS 12913-2:2018, D.4,
approximating a monaural microphone measurement.
After applying the recording equalization, the remaining signals of the left and right channels are separately
processed to determine (psycho-)acoustic metrics.
The different metrics used (e.g. L , L , L , L , N , N , N , S , IACC) shall be linked to the
Aeq,T Ceq,T AF5,T AF95,T 5 95 rmc 50
perception and the assessment of the concerned people [ISO/TS 12913-2]. In general, the consideration
of acoustic analysis results shall provide a basis for the evaluation and classification of soundscapes by
complementing the perceptual data. Moreover, based on the results of the binaural data analysis, given
sufficient position data, maps based on psychoacoustic and other data can be determined. For more
information, see Annex D.
8 Triangulation
The general idea of triangulation is to achieve a higher level of validity if different methods applied lead
to the same result and hierarchical agglomerative clustering complement each other. Triangulation for
soundscape measurement is a technique that facilitates validation of data through cross verification of three
components: people, context and acoustic environment (see Reference [16]). In particular, it refers to the
application and combination of several research methods in the study of the same phenomenon. For more
information, see Annex E.
Annex A
(informative)
Analysis of data related to Method A
A.1 General
Method A, described in ISO/TS 12913-2:2018, C.3.1, consists of rating scales containing five response
categories.
A.2 Analysis of Likert-scale data
For the analysis of the collected responses by means of Method A as described in ISO/TS 12913-2:2018, C.3.1,
numbers shall be assigned to the response categories of the category scales as follows.
The five response categories of questionnaire part 1 (see ISO/TS 12913-2:2018, Figures C.2 and C.3) from
left (‘not at all’) to right (‘dominates completely’) are assigned scale values from 1 to 5.
The five response categories of questionnaire part 2 (see ISO/TS 12913-2:2018, Figure C.4) from left
(‘strongly agree’) to right (‘strongly disagree’) are assigned scale values from 5 to 1.
The five response categories of questionnaire part 3 (see ISO/TS 12913-2:2018, Figure C.5) from left (‘very
good’) to right (‘very bad’) are assigned scale values from 5 to 1.
The five response categories of questionnaire part 4 (see ISO/TS 12913-2:2018, Figure C.6) from left (‘not at
all’) to right (‘perfectly’) are assigned scale values from 1 to 5.
For all category scales, the level of measurement is ordinal, which means that the median values should be
reported as the measure of central tendency, and the interquartile range as the measure of dispersion. See
Table A.1.
Table A.1 — Assigned scale values to rating scales of Method A and statistical measures
Scale values to be Measure of central Measure of
Part (see ISO/TS 12913-2)
assigned tendency dispersion
1 (Sound source identification) 1 ,2 ,3, 4, 5 median interquartile range
2 (Perceived affective quality) 5, 4, 3, 2, 1 median interquartile range
3 (Assessment of surrounding sound envi- 5, 4, 3, 2, 1 median interquartile range
ronment)
4 (Assessment of the appropriateness) 1, 2, 3, 4, 5 median interquartile range
NOTE ISO/TS 12913-2 refers to the questionnaire part 2 as measuring 'perceived affective quality'. However,
recent debate has arisen in the community as to whether or not the terms used are affective or cognitive attributes
(see Reference [17]). Therefore the more general term 'perceptual attribute' is preferred.
Rating data collected via questionnaires should be linked to the results of the acoustic data analyses
in order to identify potential relationships (see References [18] and [19]). These relationships may be
investigated by means of statistical analyses, such as correlation analyses, linear regression or ANOVA (see
Reference [20]). For ordinal data, the Spearman's rank correlation coefficient, r , should be calculated
spearman
(see References [21][22].
The determined correlation coefficient should be reported as a numerical measure of statistical relationship
between two variables. In addition, the statistical significance of the correlation should be determined, and
the p-value reported.
A.3 Analysis of circumplex coordinates based on perceptual attribute responses
Environmental psychologists have repeatedly demonstrated that when people are asked to freely describe
how they perceive environments they respond affectively (see Reference [23]). These affective responses
can be represented in a two-dimensional model called a circumplex, based on the inter-correlations between
attributes (see Reference [24]). While the affective nature of these responses is important for the theoretical
context (see A.2, Note), the two-dimensional circumplex arrangement of attributes is a mathematical
construct.
In the soundscape circumplex model, the main dimension is related to how pleasant or unpleasant the
environment was perceived, and therefore noted as pleasantness. The second dimension is related to
the amount of human and other activity (see Reference [23]). For soundscape, this second dimension is
represented by how eventful or uneventful the acoustic environment is perceived to be, and therefore noted
as eventfulness (see Reference [25]). An eventful environment is busy with human activity, for example a
city centre or other sound events produced by non-human agents, whereas an uneventful environment is
completely devoid of human activity, for example late evening hours in a residential area without social,
commercial and industrial activity. If the previously mentioned pleasantness and eventfulness axes are
taken as perpendicular, further labelling corresponding to human judgments can be prescribed to two
additional axes rotated 45° on the same plane. At a rotation of 45° from the two main dimensions, are two
alternative dimensions representing environments that are chaotic versus calm, and environments that
are monotonous versus vibrant (see References [25] and [26]). According to the two-dimensional model,
vibrant soundscapes are both pleasant and eventful, chaotic soundscapes are both eventful and unpleasant,
monotonous soundscape are both unpleasant and uneventful, and finally calm soundscapes are both
uneventful and pleasant.
NOTE 1 Based on the tradition in environmental noise research, the term ’annoying’ is used instead of ‘unpleasant’
in the model used in this document. However, some sociological models ascribe a stronger sense of intentionality to
sounds that are annoying compared to those that are unpleasant.
The results from ISO/TS 12913-2 (see A.1) are processed to derive the values on two dimensions
(pleasantness and eventfulness) for each response. Results can be reported in a two-dimensional scatter
plot with coordinates for the two dimensions ‘ISO Pleasantness’ and ‘ISO Eventfulness’ (see Figure A.1). The
coordinates for ISO Pleasantness are plotted on the x axis, and the coordinates for ISO Eventfulness on the y
axis. Every data point in the scatter plot represents one response collected in ISO/TS 12913-2.
The coordinate for ISO Pleasant P is calculated by means of Formula (A.1):
ISO
P = cos()θξ⋅ (A.1)
ISO ∑ ii
λ
P
i=1
The coordinate for ISO Eventful E is calculated by means of Formula (A.2):
ISO
E = sin θξ⋅ (A.2)
()
ISO ii
∑
λ
E
i=1
where i indexes each circumplex scale moving counter-clockwise around the circumplex. θ is the angle for
i
th th
the i circumplex attribute, and ξ is the Likert-scale response value for the i attribute in the range
i
15, . The standard angles are given below:
[]
θ =°0 is pleasant;
θ =°45 is vibrant;
θ =°90 is eventful;
θ =°135 is chaotic;
θ =°180
is annoying;
θ =°225 is monotonous;
θ =°270 is uneventful;
θ =°315 is calm.
In Formula (A.1) and Formula (A.2), cos()θ is used as a trigonometric projection function to adjust for the
i
45° rotation in the two-dimensional model (Figure A.1). λ provides a scaling factor to bring the range of
P and E values to []−+11, :
ISO ISO
ρ
λ = cos()θ (A.3)
P i
∑
i=1
and
ρ
λ = sin()θ (A.4)
E i
∑
i=1
where ρ is the range of the possible response values (i.e. ρ=−51=4 for the Likert scale values given in
Table A.1).
Figure A.1 — Example of graphical representation of Formula (A.1) and Formula (A.2)
NOTE 2 The angles given in Formula (A.1) and Formula (A.2) represent the idealised arrangement of the attributes
around the circumplex. Further examination has demonstrated different degrees of strict arrangements (i.e.
circumplexity) in English and across other languages which require adjustment (see Reference [28]). When applying
the formulae, it is recommended to use the adjusted angles given in Table A.2. When this is done, in order to ensure
that the scaling to -1/+1 is maintained, it is recommended that raw responses are centred at 0 by subtracting the
midpoint of the range (i.e. 3 for a 1-5 Likert scale), prior to applying the formulae.
A.3.1 Using translated attributes and adjusted angles
To ensure the accurate assessment and comparison of soundscapes across different cultural and linguistic
contexts, it is crucial to consider the potential variations in the perception of the eight attributes of the
soundscape circumplex model. The translation of the attributes to languages other than English must be
performed carefully to retain their intended meaning and facilitate cross-cultural studies. The Soundscape
Attributes Translation Project (SATP) is an international collaboration that was initiated to address this
[28]
issue . When using translated versions of the eight soundscape attributes of the circumplex, it is important
to apply adjustments of the angle θ to ensure that the attributes align correctly across languages. Research
i
has shown that different languages exhibit varying levels of confidence in retaining the quasi-circumplex
[28]
structure of the soundscape circumplex model originally proposed . For the adjusted angles to be applied
in Formula (A.1) and Formula (A.2) it is recommended that the translated attributes should have a high level
of confidence in any given language. If this is the case, assuming that individual soundscape assessment data
has been collected in compliance with ISO/TS 12913-2 and using the translated attributes reported in
Reference [28], the θ values in Table A.2 can be applied.
i
Table A.2 — Translations from the Soundscape Attributes Translation Project (SATP) in languages
other than English, with angles θ to be implemented for adjustment
i
Language Pleasant Vibrant Eventful Chaotic Annoying Monotonous Uneventful Calm
English Pleasant Vibrant Eventful Chaotic Annoying Monotonous Uneventful Calm
0° 46° 94° 138° 177° 231° 275° 340°
Standard  / يكيمانيد  / ةايحلاب ضبان / يوضوف ريغ / عتمم ريغ نم يلاخ / دماه
راس / عتمم لمم / بيتر ئداه
Arabic يويح / سمحم ثادحلأاب لفاح بخاص جعزم / راس ثادحلأا
0° 36° 45° 135° 167° 201° 242° 308°
引人注意的 /
恼人的 / 枯燥的 / 平淡无奇的 /
Chinese 热闹的 / 有故事性的 / 喧闹的 /
愉快的 烦人的 / 无趣的 / 难以共鸣的 / 平静的
(Mandarin) 有活力的 引人遐想的 / 混乱的
烦躁的 单调乏味的 无体验感的
身临其境的
0° 18° 38° 154° 171° 196° 217° 318°
uzbudljiv / sadržajan / kaotičan / besadržajan /
Croatian ugodan neugodan dosadan smirujuć
živahan pun događaja uznemirujuć prazan
0° 84° 93° 160° 173° 243° 273° 354°
aangenaam / levendig / druk / chaotisch / onaangenaam saai / rustig / kalm /
Dutch
prettig vrolijk dynamisch hectisch / onprettig eentonig statisch rustgevend
0° 43° 111° 125° 174° 257° 307° 341°
lebendig /
ereignisreich / chaotisch / störend / monoton / ereignisarm / ruhig /
German angenehm abwechslungs-
dynamisch hektisch lästig eintönig statisch erholsam
reich
0° 64° 97° 132° 182° 254° 282° 336°
χωρίς
με πολλά ή /
μονότονο / πολλά ή / και
Greek ευχάριστο ζωντανό και σημαντικά χαοτικό ενοχλητικό ήρεμο
βαρετό σημαντικά
γεγονότα
γεγονότα
0° 72° 86° 133° 161° 233° 267° 328°
menyenang-
Indonesian bersemangat ramai ribut mengganggu menjemukan sepi tenang
kan
0° 53° 104° 123° 139° 202° 284° 308°
piacevole / vivace / dinamico / caotico / spiacevole / monotono / stabile / calmo /
Italian
confortevole stimolante vario confuso irritante noioso stazionario tranquillo
0° 57° 104° 142° 170° 274° 285° 336°
유쾌한 / 활기찬 / 활동적인 / 혼란스러운 / 불쾌한 / 단조로운 / 비활동적인 / 조용한 /
Korean
기분 좋은 생동한 역동적인 혼잡한 성가신 지루한 정적인 차분한
0° 56° 90° 124° 151° 251° 275° 288°
agradable / estimulante / con actividad / caótico / desagradable monótono / sin actividad / calmado /
Spanish
placentero vibrante dinámico confuso / molesto aburrido estático tranquilo
0° 41° 103° 147° 174° 238° 279° 332°
behagligt / levande / kaotiskt / störande / enformigt / händelselöst
händelserikt / lugnt / stilla
Swedish trivsamt / spännande / rörigt / obehagligt / andefattigt / / inaktivt /
livligt / aktivt / rogivande
tilltalande uttrycksfullt bullrigt otrivsamt livlöst passivt
0° 66° 87° 146° 175° 249° 275° 335°
Turkish keyifli coşkulu hareketli karmaşık rahatsız edici tekdüze durağan sakin
0° 55° 97° 106° 157° 254° 289° 313°
Translations and related angles are reported only for languages currently exhibiting high levels of confidence as per Reference [28]. English
adjusted angles are reported too, as these would be needed for comparison with other languages in the circumplex space.

A.3.2 Statistical analysis and visualisation
For any set of responses processed as above, a vector of P , E values, which are continuous variables
ISO ISO
from -1 to +1, is created. These groups can include all responses in a given location, from a particular
demographic group, collected at a particular time of day or season, and so on. These coordinate vectors
should then be summarised by reporting summary statistics (mean and standard deviation). Circumplex
data collected via questionnaires should be linked to the results of the acoustic data analyses in order to
identify potential relationships. These relationships may be investigated by means of statistical analysis,
[29]
such as correlation analyses or ANOVA, or more advanced statistical modeling and machine learning
methods. The adequate analysis depends on the structure of the analysed dataset and the question under
investigation. Where possible, confidence intervals, p-values, and/or predictive performance should be
reported. The recommended measures of predictive performance for such models are R (including adjusted
2 2
R and marginal/conditional R , when relevant for the model type) and root mean squared error (RMSE),
reported according to the [-1, +1] scaled circumplex coordinates.
Three visualisation variations are recommended for circumplex coordinate data:
a) scatter plot of the calculated coordinates (see Figure A.2 a));
b) Kernel Density Estimation (KDE) plots with decile contours or continuous colourmaps (see
Figure A.2 b));
c) Simplified KDE plots showing the median contour (see Figure A.2 c)).
WARNING — Kernel Density Estimates are unreliable for low sample sizes. Density plots as shown in
Figure A.2 b) and Figure A.2 c) should be avoided for sample sizes N < 30, in which case scatter plots
are the preferred visualisation method.
a) Example of a scatter plot of circumplex coordinates

b) Example of a decile KDE distribution plot of circumplex coordinates for one location

c) Example of simplified median contour distribution plots comparing two locations
Key
LocationID: RussellSq
LocationID: RussellSq
LocationID: CamdenTown
Figure A.2 — Demonstrations of the visualisation methods for circumplex coordinate data
EXAMPLE Implementations of the analysis and visualisation methods shown in this Technical Specification have
[30]
been implemented in publicly available software. See for example, the Soundscapy package which provides an
implementation in Python.
Annex B
(informative)
Analysis of data related to Method B
B.1 General
Method B, described in ISO/TS 12913-2:2018, C.3.2, consists of five-point unipolar continuous rating scales,
a ranking test and free words data, which should be separately analysed. The free words data refer to
method B as described in ISO/TS 12913-2:2018, C.3.2, where, after listening, the soundwalk participants are
requested to write down their thoughts and feelings while still at the specific location.
B.2 Determination of central tendencies of responses
Part 1 (assessment of the sound environment —
...