J. Service Science & Management, 2010, 3, 429-439
doi:10.4236/jssm.2010.34049 Published Online December 2010 (http://www.SciRP.org/journal/jssm)
429
Urban Soundscape Informational Quantization:
Validation Using a Comparative Approach
Philippe Woloszyn1, Thomas Leduc2
1ESO Lab., Université de Haute Bretagne and CNRS, Rennes, France; 2CERMA Lab., ENSA Nantes and CNRS, Nantes, France.
Received September 1st, 2010; revised October 12th, 2010; accepted November 17th, 2010.
ABSTRACT
Through interaction with environmental parameters such as light or sound, urban and architectural spaces generate
ambiences with identifiab le characteristics. This notion of ambiences is related to the human being position through its
perception o f environmental physica l phenomenon du ring a pedestrian walk. Presented work aims to evaluate, so as to
characterize, the impact of sound ambiences (soundscape) onto an urban pedestrian pathway using GIS spatial dy-
namical mapping. To carry out this scheme, our research work within AMBIOFLUX project concerns spatial interac-
tion between sound ambience (soundscape) and man urban spatial trajectory (soundwalk). Spatial impression of
soundsources or soundmarks has to be both defined through acoustical measurement and perception informational
evaluation. The remainder of this paper is dedicated to the evaluation’s methodology of the pedestrian pathway’s
acoustic fingerprint using the GearScape spatial formalism described thereafter. Preliminary results we have obtained
will also be presented and validated.
Keywords: Ambience, Soundscape, Soundmarks, Entropy evaluation, Information dimensioning, GIS semantics
1. Ambient Environment Perception and
Representation
AMBIOFLUX project concerns spatial interaction be-
tween sound ambience (soundscape) and a human urban
spatial trajectory (soundwalk). It defines ambiences as an
anthropocentric view of the global environmental pro-
duction through physical, human and built constraints of
architectural and urban design. To model it, we consider
urban space as a field of data aimed at ambiences physi-
cal parameters description through multi-phenomenal
characterization [1].
An application of this principle is currently proceeded
through this interdisciplinary research project, called
AMBIOFLUX, funded by CNRS, French National Cen-
tre for Scientific Research, and MEEDDAT Ministry
(French Ministry dedicated to Ecology, Energy, Sustain-
able Development and land use planning) under PIRVE’s
contract (Programme Interdisciplinaire de Recherche
Ville et Environnement). The corresponding research
work aims at producing dynamical urban environmental
index for spatial interaction indicators between ambience
and urban walkthrough [2].
From a software point of view, GearScape (a customi-
zation of the OrbisGIS project initially developed by E.
Bocher, F. González Cortés and T. Leduc in the CNRS
FR 2488 context [3]) original spatial formalism process-
ing aims to qualify pedestrian spatial interactions with
producing a set of ambient dynamical indicators.
This approach is therefore founded upon elementary
sound sources description, organization and recognition
with proceeding to its elementary hierarchic identifica-
tion and systemic modeling [4,5].
2. Soundmarks as Soundscape
Psychophysical Encoded Elements
Soundscape informational dimensioning will be consid-
ered through urban sound sources spatial psychophysical
indicators formulation, soundmarks. Murray Schäfer in-
troduces the word “soundmarks” as a derivation of the
word “landmark”, to identify sounds which sign the out-
standing role of sounds to characterize a place [6]. In this
sense, soundmarks describe sound events which get a
specific informational status, mainly deno tative, that me-
ans they are strong identity revealers.
The dedicated informational order scaling corresponds
to near-order indices, which can be evaluated through
maximal information entropy. Therefore, soundmarks are

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 430
defined as maximum-entropy sources a soundwalker can
meet, defined as the most consciously emerging urban-
situated events within his world-line, for a given urban
trajectory. They are computed using local entropy sour-
ces calculation H [7], as described in Equation (2).
2.1. Worldline as a Soundwalk Representation
Model
For the pedestrian who wanders around the urban space,
the ambient phenomenon overlapped with the ur- ban
landscape might be considered as a marker of the entire
phenomenon distributed around a place, creating a per-
ceptible atmosphere for anybody located in this space
[8].
The fundamental principle of the world-line considers
the temporal structure of perception, claiming that an
observer identifies the beginning and the end of a per-
ceived event. This assumption states that the observer
codes its corresponding time-segment, or world-line, as a
causal attribute of the perceived event [10,11]. In our
case, for a given subject and within a given observation
period, soundscape knowledge of an observer is relevant
to sound sources emergence and occurrence frequency.
Those two subjective characteristics constitute the main
scaling dimensions which have to be defined for infor-
mational quantization .
2.2. Entropy Index Dimensioning
Among various variables, the entropy is the thermo dy-
namical simplest quantity to be applied to non-physical
systems, as it is considered to be a measure of system
disorder within informational datasets.
Unlike thermodynamic entropy, being a “content-full”
concept specific to thermodynamic systems, statistical
entropy applied here qualifies informational probability
distribution as a “content-free” syntactic concept, a quan-
tity calculated from the numerical properties of the “vir-
tual system” distribution laws.
It is important to note that even Boltzmann’s view of
the second law of thermodynamics, using the entropy
term [12] as a law of disorder into an open system, con-
firms this “content-free” ontological status of statistical
entropy [1 3].
Following this assumption, the challenge of the work
pioneered by Shannon and Jaynes [14,15] was to extend
the entropy concept and to apply its measure in as many
different contexts as possible.
Therefore, Shannon’s information theory [14] together
with E.T. Jaynes principle of Maximal Entr opy [15] pro-
vides a constructive criterion for setting up probability
distributions on the basis of partial knowledge. This cri-
terion leads to a statistical inference model called maxi-
mum-en tropy estimate.
2.3. Application: Soundscape Entropy
Quantification
Sound source geo-localization (xy location into the urban
maze), is gathered here with two spatial extends: sound
pressure level and soundmark entropy values.
Equivalent Sound Level is formulated in terms of the
equivalent steady noise level which in a stated period of
time would contain the same noise energy as the time
varying noise during the same time period [16]:
10
1
10log10 eqi
L
eq i
i
LT
T

 


 (1)
The second indicator corresponds to Shannon entropy
calculation [14] as:
 
1
log
xX
Hpx
px

 (2)
which describes the uncertainty quantity by the informa-
tion which we do not have about the state occupied by
the concerned source.
Probability p(x) is based on empirical frequencies
measurement issued from observation statistics from
inquiries and so undwalkers expr essions [17], and is actu-
ally calculated from the frequency occurrence of the re-
lated event x within the recorded soundscape; Huff-
mann’s perceptual encoding can then be enacted through
the relevant soundscape Zipf-Pareto law for the sources
distribution [18].
The resulting quantity is a measure of the uncertainty
of the soundscape events occurrence: the higher the H
value, the more unpredictable the constitutive sound
events; in other words, entropy index H constitutes a
reliable soundscape originality measurement.
In our case, Maximum Entropy principle considering n
sound sources as discrete random variables xi with en-
tropy Hi, the “largest remaining uncertainty probability
distribution” can be estimated from the 2n-1 dimensional
vector, called entropy vector E(x), following equation (3)
formulation:
  
1
1
1log, ,
n
in
i
Expx MaxHH
n
 
 (3)
This quantity helps to study the behavior of the in-
formation taking into account the n sources composing
the studied soundscape. The related quantity E(x) gives
us the extension value of the corresponding soundmark,
considering its ability to reach the listener (source
emergence value) during the corresponding urban
soundwalk.
For a given soundscape, sensation scaling proceeds

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 431
fr
Firstlish texts word occurrence fre-
f’s law may be stated mathematically as:
om “soundmarks” (maximum entropy sound event) to
“Keynotes” (null-en tropy background noise) [1 9].
2.4. Zipf-Pareto Law Sources Distribution
Dimensioning
applied for Eng
quency determination, empirical law known as “Zipf
law”, named for Harvard linguistic professor George
Kingsley Zipf, models the occurrence of distinct objects
in particular collections [20 -23]. Zipf law says that the ith
most frequent object will appear 1/ ith times the fre-
quency of the most frequent object in the collection.
Moreover an expression of universal regu larities, this law
is applied in numerous domains: in English texts word
occurrence frequency [20,24,25], as well as populations
of cities [26-28], immune system characterization [29],
bibliographical classification or prediction [30,31], or
cancer classification [32]. Nevertheless, except musical
[33-35] and audio medical signal [36] applications, we
did not find Zipf law application for other audio domains
such as soundscape acknowledgment in scientific litera-
ture.
Zip
 
log log
x
f
Cs k (4)
where fx, the frequency of the unit (word form or lemma)
dscape application, mathematical expres-
si
m Geographical
are purely of vec-
to
ustrated in Figure 1, the global spatial process
w
d one develops a ra-
th
e will successively apply and detail all
to input data already mentioned, 7 sound-
m
qualitative analysis consists
in
k-order (Zipf) analysis has been proc-
es
in Nantes historic heart, a
having the rank k, s, the exponent coefficient (near to 1
for French language word frequency distribution), and C,
a constant.
For soun
on of Zipf law involves the number of occurrences of
a done sound source, understood as an acoustic
emerging event. Within a given soundscape, relation-
ship between the constitutive sound sources emegence
with respect to their occurrence frequencies should then
provide a rank-order Zipf power law, with a specified
entropy dependent slope. The resulting event density
probability distribution will then provide information
quantification through entropy indexing, thanks to the
use of GearScape dedicat ed s p a t i a l s e ma n t i c s y s t e m .
3. Soundscape Geoprocessing
The main idea here is to take benefit fro
Information well-known concepts and techniques and
apply them both to the sp atial interactions between sou nd
ambiances and an urban pedestrian walk. Therefore, we
have to map soundmarks effects onto the pedestrian
pathways and compute some relevant indicators to char-
acterize the environmental interaction process. Among
all of them, we have decided to focus on the spatial
sound pressure integrated levels and the entropy index.
The main add-on of this paper is clearly to couple
soundscape emergence concept with a pedestrian mobil-
ity, that is to say a dynamic process. Confluence is
achieved in the context of a specific Geographical In-
formation System called GearScape.
The data that have to be processed
r type. They are provided by the French IGN agency
(lay- ers extracted from the so called BD ORTHO® spa-
tial database). The study areas have been selected be-
cause of their wide morphological variabilities concern-
ing both the urban fabric and the corresponding net-
works.
As ill
e have designed concerning the Strasbourg use case is
divided into 10 main steps combining some well-known
OGC [38,39] functions. It consists in a sort of raster ap-
proach, which relies on an orthogonal regular grid based
layer produced using an operator developed in the con-
text of the UrbSAT plugin [40,41].
Unlike to the 1st approach, the 2n
er different method (see Figure 2). Instead of mesh-
ing the surrounding soundscape it discretizes the sound-
walks themselves. This processing schema is much more
robust and efficient. This is the one we have adopted
with the Nantes town centre use case.
4. Comparative Case Studies
4.1. Contexts
In this section w
the processes presented in Section 3 to one of the main
study area of the AMBIOFLUX project in Strasbourg
suburbs and to the Nantes city center. First case study is
located in the north suburbs of the French city S tr a s b ou rg
(see Figure 3). It corresponds to a rectangular area of
less than 2.9 km, from north to south, by 2.2 km, from
west to east. In this region of interest, 5 different pedes-
trian pathways are defined with an average length of 4.7
km and a standard deviation equals to 1.2 km. All those
pathways connect Schiltigheim city to Strasbourg’s rail-
way station.
In addition
arks have been defined. In all corresponding locations,
sound recording have been performed and analyzed for
the 5 more significant ones.
After recording operation,
operating a multi-sources description of the whole
sequence. A statistic of the resulting description items
will then provide their respective o ccurrence frequencies,
in order to be plotted regarding their corresponding
emergence levels.
Thereafter, a ran
sed taking each constitutive source within the sound
marked sequence into account.
Second case study is located

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 432
Figure 1. The processing schema we have adopted in the Strasbourg use case. The sequence is composedf 10 main opera- o
tions. Input maps are 45 degrees wide hatched, intermediate results have no background color and final output results are
colored in gray.
Figure 2. The processing schema we have adopted in the Nantes use case. The sequence is composed of main operations.
est-coast located city in France (see Figure 4). Walks
lts and Discussion
nk-order slope values of
of the recorded points clearly discriminates poor sound-
ed
6
Input maps are 45 degrees wide hatched, interme diate results have no background color and final output results are colored
in gray.
w
were led the same day evening, from town-hall to Roo-
sevelt Court. The fourteen fixed recording sequences
points have then been analyzed to compute Zipf rankor-
der analysis and calculate the corresponding entropy
values. The high dens ity of analyzed points will enable a
fine spatial discretization for soundmarks entropy evalu-
ation, according to the perceived sources occurrence fre-
quencies.
4.2. Resu
As illustrated by the low Zipf ra
the Figure 5, emergence density probability distribution
scapes (“low-fidelity” soundscapes in the sense of M.
Schäfer [6,19], corresponding here to points 1 (Rempart)
and 3 (Autoroute de l’Est)), from rich ones (“high-fidel-
ity” soundscapes, composed with hierarchized numerous
sound sources, corresponding here to points 5 (Cité Nu-
cléaire), 6 (Schiltigheim) and 7 (Z. A. Mittelfeld)). As a
result, we obtain a set of spatial punctual positions and,
for each of them, ambient indicators such as the aggre-
gated Leq, coupled together with entropy evaluation. The
numerical values we obtain are presented in Figure 7.
To characterize each studied pedestr ian pathways, some
ambient indicators are produced such as the aggregat

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 433
Figure 3. Zoom on the Region of Interest located ine
north-west of the French city Strasbourg. All pedesian
th
tr
pathways (dotted poly lines) share the same origin (disc in a
square, north of the map) and destination (disc in a square,
south of the map) points. The seven soundmarks emergence
areas are numbered and represented by pale gray concen-
tric discs.
Figure 4. Zoom on the Region of Interest located in the
town center of the French city Nantes. All pedestrian pa- th
ways share the same origin (north of the map) and destina-
tion (south of the map) points. The 14 soundmarks emer-
gence areas are numbered and represented by concentric
discs.
Figure 5. Zipf rank-order laws, entropy values and sound
levels for soundwalk points in Strasbourg suburbs.
Leq or the maximum entropy value all along the path-
way.
Concerning Nantes’s walkthrough area, calculated en-
tropy values (see Figure 6) clearly discriminate three
sets of soundmarks: a “low entropy group”, scaling val-
ues from 0.6 to 0.7 (points 3,4,8 and 15), a “middle” one,
with entropy values from 0.7 to 0.8 (points 1,2,6,7, and
17) and a “high entropy group” gathering values over 0.8
(points 16,18,19,20). One can note that this values dis-
tribution is confirmed by the so undmarks Zipf rank-order
laws, which regression values scales from 0.02 to 0.29
for the first group, from 0.2 to 0.5 for the second one,
and from 0.4 to 0.5 and more for the “highest entropy
group”.
Another remark concerns the relatively lower entropy
values bracket for Nantes soundmarks, compared to
Strasbourg’s ones. This fact can be explained by the dif-
ference of areas extensions: since Strasbourg’s area,
dedicated to seven urban soundwalks analysis is about 6
km², Nantes area study gather fourteen points in about
only 0.7 km². Consequently, high soundmarks density
within this last area, concentrated in a relatively homo-
geneous urban district, can not offer the same sound am-
bience variations than the first one in Strasbourg.
As may be noticed, paths number 2 to 5 (see Figure 3)

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 434
Figure 7. Time-line approach of the equivalent sound level
and the entropy index of Strasbourg subur bs.
share the same maximum entropy value. It is due to the
fact that they all come across the same predominant
soundmark (south-most one, tagged as “Rempart”).
What seems important to notice here is that both en-
tropy and equivalent sound level signatures do not fully
match in the particular case of the 5th pathway (see Fig-
ure 7). Indeed, crossing half the first soundmark emer-
gence area (south-most one, tagged as “Rempart”), the
“soundwalker” faces a particular acoustic event that is
not significant in term of noise energy but in term of in-
formational content (entropy). This clearly shows that
entropy index is not strictly correlated with the equiva-
lent sound level index.
For Nantes city centre area, informational treatment
exposed in Figure 8 provides a time-sliced discreet qua-
ntitative information on the soundscape originality and
intelligibility o f the concern ed sound environment dur ing
the walks. Thus, the “distance gap” between sound marks
(observation points) has to be well-dimensioned to obtain
a continuous varying entrop y signal. As observed here, a
relatively dense soundmarks discretization within the ur-
ban space allows a quasi-coherent entropy signal along
the soundwalk pathways.
5. Validation Method: Entropy vs.
Soundscape Multi-Sources Combination
5.1. Reference Model Presentation
In order to evaluate the stability of our method, we have
decided to compare the results we obtained on the city of
Nantes with an already published soundscape quality
map. This map was aimed to produce sonic ambiance’s
compositions during a soundwalk, constitutes a unique
attempt of sound ambiance qualitative analysis in scien-
tific literature by Léobon in the 90’s [45]. This structural
ignatures
and phenomenological approach of Nantes city sound-
scape is based on environmental sources sonic s
Figure 6. Zipf rank-order laws, entropy values and sound
levels for soundwalk points in Nantes city center.

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 435
Figure 8. Time-line approach of the equivalent sound lev
d the entropy index of Nantes city center (The 1el anst
orange colored,
fully the districts’ various sonic atmospheres. Sampling
of this soundwalk is relevant to an exploration process,
discretizing urban space through obvious soundscape
changes.
Each record ed sequence, examin ed thro ugh headphon e
listening, is transcribed into a list of sonic ite ms, grouped
together according to sound sources families, thus con-
stituting a hierarchic structure between three extreme
uses of public spaces (as shown in Figure 9): the pedes-
trian sequence function (“Présence” in blue), the traffic
line function (“Activité Mécanique” in red), and the ani-
mated places (“Animation” in yellow).
The cartographic mapping we reefers to associates a
dedicated color to the corresponding area for each re-
cording point within the soundwalks, according to the
soundscape multi-sources composition. The ten colours
used to represent the various sonic atmospheres of a city
centre are the following (see Figure 10):
1) Blue: pedestrian, open spaces or residential sonic
areas;
2) Purple: open spaces sonic areas with traffic noise in
the background;
3) Green: mixed sonic areas with dominant anthropo-
genic noise;
4) Light green: mixed sonic areas with predominant
anthropogenic noise, moderately animated;
5) Yellow: intensely animated sonic areas;
6) Salmon pink: mixed sonic areas with pedestrian and
road traffic, without sonic signs of activity;
7) Light yellow: mixed sonic areas with pedestrian and
road traffic, moderately animated;
8) Orange: mixed sonic areas with dominant road traf-
fic noise;
9) Ochre: mixed sonic areas with dominant road traffic
noise, animated;
10) Red: predominant road traffic noise.
soundwalk is blue colored, the 2nd one is
the 3rd one is red colored).
analysis.
In this approach, “environmental sound sources”
composing our perception of the urban landscape are
translated into descriptive items, which statistical ana-
lysis leads to a soundscape multisources cartography.
This reference qualitative sonic inventory is fed with
urban soundwalks, marked out by relevant recording
points, short sound sequences representing rather faith-
Animation
présence Acticité Mécanique
70 50 30
Figure 9. Sound sources color trade-off (extracted from
[45]).
50 60
30 80

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 436
Figure 10. The ten colors used to represent the various
5.2. Ambient Multi-Sources Inform
sonic atmospheres of a city centre (extracted from [45]).
Th ith strictly the same points used for the
re
mputing)
and am
methodology [47]).
The reference cartographic representation uses the
previously described colors to indicate the composite
multi-sources areas, as seen on the following sound am-
bience map, based on Léobon’s summer evening sound-
walks in Nantes city center (see Figure 11) [48].
Despite their short extends, those paths reveal various
atmospheres that a passer-by would face when wandering
in the urban maze. As a consequence of this case study
limited area, the great density of analyzed points enables
a very fine qualitative discretization of the paths for mi-
cro-structural an alysis of the soundwalks.
5.3. Results Comparison: Entropy Values vs.
Ambiance Multisource Qualification
In ord
ined all along the three pathways, we will compare
pre-
sented ambiences cartography Figure 11
(mixed areas with dominant road traffic noise) colors.
Their respective entropy values scale from 0.61 to 0.75
for all the concerned points (points 1, 3, 4, 5, 8, and 15).
A common characteristic of all those punctual positions
is their high road traffic noise component, which tends to
“annihilate” the other sound sources, and impoverish th
ational
Validation
e Nantes entropy and Zipf rank-order law calculations
were produced w
ference sound ambience cartography we will present
now. As we were part of this research [46], we used the
sa
e
me audio samples to provide the previous inform-
ational computing: this enables direct comparisons be-
tween originality measurement (informational co
bient multi-sources characterization (Léobon’s
er to validate the maximum entropy values ob-
ta
them to the soundscape ambient multi-sources data,
d in the soun
above. To facilitate this data set comparison, we have
reused th e color code alr eady presen ted. Thus, in th e Ta-
ble 1, each row color corresponds to the characteristic of
the corresponding soundmark.
In this case study, poor sou ndscapes correspond to red
(areas with predominant road traffic noise) and orange
Figure 11. The Nantes city center sound ambiences on
summer evening [48], including the 3 soundw alks paths and
the recording punctual positions.

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 437
Ta
N LEQ dB(A)
ble 1. Soundmarks attributes (Nantes’ use case). Com-
parative table between the soundmarks informational
characterization, multisource composition and sound levels.
Color coding is refeering to the sound ambience cartog-
raphy (see Figure 11).
PT SLOPE ENTROPY AMBIANCE
DESCRIPTIO
1 0.41 0.74 Pedestrian and road
traffic 70
2 0.31 0.75 traffic and animated 75
3 0.02 0.61 Predominating road
traffic noise 75
4 0.11 0.62 Predominating road
traffic noise 80
5 0.2 0.72
Mixed with dominant
road traffic 75
6 0.5 0.78
Mixed with dominant
human noise 70
7 0.47 0.79 Pedestrian and road
Pedestrian, road
traffic
8 0.29 0.67 Mixed with dominant
road traffic 75
15 0.11 0.65 Mixed with dominant
road traffic 70
16 0.47 0.81 Pedestrian, road
traffic and animated 75
17 0.35 0.78 Pedestrian and road
traffic 75
18 0.45 0.8 Pedestrian and road
traffic 60
75
19 0.52 0.81 ixed with dominant
human noise 65
20 0.41 0.85 Pedestrian, road
traffic and animated 70
M
corresponding soundscapes. Moreover, those points are
characterized with high sound pressure levels (around
75-80 dB(A)).
On the other side, soundscape richness is signed up
with higher entropy values: the corresponding punctual
positions display entropy values from 0.75 to 0.85
(points 2, 6, 7, 16, 17, 18, 19, and 20), signing more “an-
thropogenic” or “animated” soundscapes, as “pedestrian,
road traffic and animated areas” (light yellow color
coded), or “mixed areas with dominant human noise”
(light green). Sound level values of those points are
mostly lower (around 65-70 dB(A)).
5.4. Discussion
Considering this sharp fitting between soundscape mul-
tisource characterization and their entropy evaluation,
this approach allows to characterize soundscape original-
ity in terms of “phonicity”, measuring the intelligibility
of a soundscape, which is our ability to identify all its
components. Moreover, we can notice that a noise level’s
map would fail to provide information as accurate about
a district’s sonic identity. The comparison between noise
oth vehicle’s traffic and
s of soundurce co-
pion csou hi
wh hu mindy en o
sc, caraduc Ziprder power law
xt f ourk wy signal
shape, more precisely its “rem ent, which
means temofect”marks. This will
enle tosider man inta-
neous” process (actually traduared en-
tro sigen tteneng the sounrk
influencec), wonsiy dy”
when the soundwalker leav influence
zone. Coation is ded frohe
Zianer polaw,/f nd
source density distributionlculate a
“dmmergewo prde
a soundwontininfoation.
7. Acknowledgements
The AMLUX ect wRS, Fch
National ter for ScientifiEDDAT
Mstrych M de, E,
Sustnablevelopm ananning) uder
PIRVE’s cPramire de Re-
herche Ville et Environnemen t).
onzález Cortés and E. Bocher, “OrbisGIS:
and entropy lev els of the same sonic walk shows without
contest that only this last can be able to discri minate very
close sonic ambiences, taking b
pedestrian activity into account.
6. Conclusions
Reference map scales the gradual intensity variation of a
sound quality in termscape multisom
erarchy,
rding
osit. Thisomposite nd qualitative
ich
ale man n be t ma ed byxpress through a f rank-oer .
Nestep o worill study the entrop
anence” compon
of soundhe “mry ef
ab consoundrk effect as a “nostan
uced through a sq
r is enteripynal whhe lisdma
disith cdering an “entropeca
es the soundmark
ecay will be statmputof thm t
pf rk-ordwer considered as a 1sou
, in order to ca
er spectrum”, able tsounark ence poovi
alk cuous rmational evalu
BIOF
Cen projas funded by CN c Reseren arch, an d ME
dicated to Ecologyini (Freninistrynergy
aie D
ontract (ent ogr d land use pl me Interdisciplinan
cPart of the GearScape software development was
funded by the AMBIOFLUX project.
REFERENCES
[1] P. Woloszyn, “From Fractal Techniques to Subjective
Quantification,” In: Landscape and Architectural Model-
ling Symposium Proceedings, Sousse, pp. 1-6, 2002.
[2] P. Woloszyn and D. Follut, “The Visualisation of the
Urban Ambients Parameters,” Computer Science for En-
vironmental Protection, Gesellschaft für Informatik (GI),
Bonn. Marburg (DE): Metropolis Verlag, 2000, pp. 173-
186.
[3] T. Leduc, F. G
a GIS for Scientific Simulation,” In: RMLL’2007, Amiens,
France, 2007.
[4] J. J. Gibson, “The Ecological Approach to Visual Percep-
tion,” Hillsdale, New Jersey, 1986.
[5] K. Richardson, “Hyperstructure in Brain and Cognition,”
psyc.99.10.031, 1999.
[6] R. M. Schäfer, “The Tuning of the World,” Knopf, New
York Republished in The Soundscape: Our Sonic Envi-
ronment and the Tuning of the World, Rochester, Ver-
mont: Destiny Books, 1977.

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 438
[7] P. Woloszyn and G. Bourdin, “Urban HyperScape: A
Community Game for Territorial Knowledge,” In: Second
International Annual Conference of Territorial Intelli-
gence, Huelva (Spain) October 24th-27th, 2007, Obser-
Empleo, Huelva, pp. 219-242.
l The-
,” University of Illinois Pre
[17] P. Woloszyn, “An Acoustic Ambience Study by Immer-
sive Sound Reonference Bu
with Sounds, Par
f the
ledge), 1992.
rsity Press, 1932.
al of Documentation
Place of Zipf’s and Pareto’s Law
Vol. 13, No. 5-6, 1988, pp.
Informa-
3, pp. 717-720.
and Urban Economics
ystem,” Biosystems, Vol. 39, No.
rowicz, “The Theoretical Foundation of Zipf’s
w in Importance of Genes
uters: MIDI-Encoded Music and the
ic Journal, Vol. 29, No. 1,
and the
vatorio Local de
[8] P. Woloszyn and G. Bourdin, “The Hyperscape Project:
Participative Game Informational Construction,” 6th An-
nual International Conference of Territorial Intelligence,
Besançon, 2008.
[9] A. A. Moles, “Théorie Structurale de la Communication
et Société,”Paris, Masson/CNET/ENST, 1988, p. 295. htt
[10] R. G. Barker, “Structure of the Stream of Behavior,” Pr-
oceedings of the 15th International Congress of Psychol-
ogy, 1957, pp. 155-156.
[11] R. G. Barker, “The Stream of Behavior as an Empirical
Problem,” The Stream of Behavior, New York, Appleton
of Century Crofts, 1963, pp. 1-22.
[12] D. A. Salvo, “Boltzmann and Hertz on the Bild- concep-
tion of Physical Theory,” History of Science, Vol. 28, No.
79, 1990, pp. 380-398.
[13] K. D. Bailey, “Social Entropy Theory: An Application of
Nonequilibrium Thermodynamics in Human Ecology,”
Advances in Human Ecology, Vol. 2, 1993, pp. 133-161.
[14] C. E. Shannon and W. Weaver “The Mathematica
ory of Information Urbana
1949. ss, a
[15] E. T. Jaynes, “Where do We Stand on Maximum En-
tropy?” The Maximum Entropy Formalism, Raphael D.
Levine and Myron Tribus, editors, The MIT Press, Cam-
bridge, MA, 1979, pp. 15-118.
[16] EPA (U. S. Environmental Protection Agency), “Infor-
mation on Levels of Environmental Noise Requisite to
Protect Public Health and Welfare with an Adequate
Margin of Safety,” EPA/ONAC 550/974004 Washington,
D.C., 1974.
cognition,” European C
is, March 2005. ilding Zipf-Mandelbrot Law,” Proceedings of the IEEE South- eastCon 2002, New York: Institute of Electrical and
Electronics Engineers, 2002, pp. 52-57.
[34] B. Manaris and Alii, “Zipf’s Law, Music Classification,
and Aesthetics,” Computer Mus
http://rp.urbanisme.equipe-ment.gouv.fr/puca/agenda/DA
FAG_WoloszynA_170305.pdf
[18] G. K. Zipf, “Human Behaviour and the Principle o
Least Effort,” Addison-Wesley, Reading, 1949.
[19] R. M. Schafer, “Music, Non-Music and the Soundscape,”
Companion to Contemporary Musical Thought, Vol. 1,
Paynter, J. Howell, T. R. Orton and P. Seymour, (ed.’s)
(Rout
[20] G. K. Zipf, “Selected Studies of the Principle of Relative
Frequencies of Language,” Cambridge, Massachusetts:
Harvard Unive
[21] B. C. Brookes, “The Derivation and Application of the
Bradford-Zipf Distribution,” Journ ,
ti
Vol. 24, No. 4, 1968, pp. 247-265.
[22] L. Egghe, “The Exact
amongst the Classical Information Laws,” Scientometrics,
Vol. 20, No. 1, 1991, pp. 93-106.
[23] M. Kunz, M. Lotka and G. K. Zipf “Paper Dragons with
Fuzzy Tails,” Scientometrics,
289-297.
[24] D. C. Blair, “Language and Representation in Information
Retrieval,” Amsterdam, Elsevier, 1990.
[25] W. Li, “Random Texts Exhibit Zipf’s-Law-Like Word
Frequency Distribution,” IEEE Transactions on
tion Theory, Vol. 38, No. 6, 1992, pp. 1842-1845.
[26] X. Gabaix, “Zips’s Law for Cities: An Explanation,”
Quarterly Journal of Economics, Vol. 114, No. 3, 1999,
pp. 739-767.
p://econ-www.mit.edu/faculty/download_pdf.php?id=5
30.
[27] R. Kali, “The City as a Giant Component: A Random
Graph Approach to Zipf’s Law,” Applied Economics Let-
ters, Vol. 10, No. 11, 200
[28] K. T. Soo, “Zipf’s Law for Cities: A Cross Country In-
vestigation, Mimeo,” An Updated Version has been Pub-
lished in 2005 in Regional Science
Vol. 35, No. 3, 2002, pp. 239-263.
[29] J. D. Burgos and P. Moreno-Tovar, “Zipf-Scaling Be-
havior in the Immune S
3, 1996, pp. 227-232.
[30] R. A. Fairthorne, “Empirical Hyperbolic Distributions
(Bradford-Zipf-Mandelbrot) for Bibliometric Description
nd Prediction,” Journal of Documentation, Vol. 25, No.
4, 1969, pp. 319-343.
[31] J. Fedo
Law and its Application to the Bibliographic Database
Environment,” Journal of the American Society for In-
formation Science, Vol. 33, No. 5, 1982, pp. 285-293.
[32] W. Li and Y. Yang, “Zipf’s La
for Cancer Classification Using Microarray Data,” Jour-
nal of Theoretical Biology, 2002, pp. 539-551.
[33] B. Manaris, T. Purewal and C. McCormick, “Progress
Towards Recognizing and Classifying Beautiful Music
with Comp
2005, pp. 55-69.
[35] B. Manaris et al., “Evolutionary Music
Zipf-Mandelbrot Law: Progress towards Developing Fit-
ness Functions for Pleasant Music,” Proceedings of Evo-
MUSART2003-1st European Workshop on Evolutionary
Music and Art, Berlin: Springer-Verlag, 2003, pp. 522-
534.
[36] E. Dellandréa, P. Makris and N. Vincent, “Zipf analysis
of Audio Signals,” Fractals Vol. 12, No. 1, World Scien-
fic Publishing, Singapore, 2004, pp.73-85.
[37] B. Mandelbrot, “Information Theory and Psycholinguis-
tics,” R. C. Oldfield and J. C. Marchall, Eds., Language,
Penguin Books, 1968.
[38] J. R. Herring, “OpenGIS® Implementation Specification
for Geographic information - Simple Feature Access-Part

Urban Soundscape Informational Quantization: Validation Using a Comparative Approach 439
eriza-
008.
Spatial SQL to Urban Indicator Definition and Pro-
formation Technolo-
zalez Cortés,
ity Global
e L’Acoustique et la Ville,
1: Common architecture,” 2006.
http://www.opengeospatial.org/standards/sfs
J. R. Herr[39] ing, “OpenGIS® Implementation Specification
for Geographic Information - Simple Feature Access -
Part 2: SQL option,” October 2006.
http://www.opengeospatial.org/standards/sfs
[40] N. Long, E. Bocher, T. Leduc and G. Moreau. “Sensitiv-
ity of Spatial Indicators for Urban Terrain Charact
tion,” In: IEEE International Geoscience & Remote
Sensing Symposium, IGARSS-2008, Boston, Massachu-
setts, U.S.A., 2
[41] E. Bocher, T. Leduc and F. González Cortés, “UrbSAT:
from
duction,” In: Free and Open Source Software for Geo-
spatial Conference, FOSS4G’2008, Cape Town, South
Africa, 2008.
[42] T. Leduc, E. Bocher and G. Moreau, “GDMS-R: A
Mixed SQL to Manage Raster and Vector Data,” In: J.
Horak, L. Halounova, D. Kusendova, P. Rapant, and V.
Vozenlek, Eds., Advances in Geoin
gies, Chapter 4, pp. 43-56. VSB-Technical University of
Ostrava, Czech Republic, 2009.
[43] E. Bocher, T. Leduc, G. Moreau and F. Gon
“GDMS: an Abstraction Layer to Enhance Spatial Data
Infrastructures Usability,” In: 11th AGILE International
Conference on Geographic Information Science, AGI-
LE’2008, Girona, Spain, 2008.
[44] T. Leduc, E. Bocher, F. Gonzalez Cortes and G. Moreau.
“GDMS: A Mixed Spatial Semantic to Enhance Urban
Knowledge,” In: 1st ECN and Keio Univers
COE Joint Workshop, Nantes, France, 2008.
[45] A. Léobon, “La Qualification des Ambiances Sonores
Urbaines”, Revue Natures - Sciences - Sociétés. Vol. III,
No. 1, pp. 26-41, 1995.
[46] A. Léobon, P. Woloszyn, M. Demais and E. Pasquier,
A-V. Blain, “Identité Sonore et Qualité de vie en Centre
Ville (Quartier Graslin, Nantes),” Rapport de Recherche
SRETIE, Ministère de l’Environnement, pp. 124, 1994.
[47] A. Léobon, “Analyse Sensorielle d’un Environnement
Sonore Urbain,” Colloqu
EAPLD, pp. 30-34, 1991.
[48] A. Léobon, “Qualification et Cartographie des Ambiances
Sonores du Centre Historique Nantais,” Rapport de Re-
cherche Ville de Nantes, p. 186, 1995.