Escolar Documentos
Profissional Documentos
Cultura Documentos
ABSTRACT
The present study reports results from an experiment that is part of
Soundscape Emotion Responses (SSER) study. We investigated the
interaction between psychological and acoustic features in the
perception of soundscapes. Participant features were estimated with
the Ten-Item Personality Index (Gosling et al. 2003) and the Profile
of Mood State for Adults (Terry et al. 1999, 2005), and acoustic
features with computational tools such as MIRtoolbox (Lartillot
2011).
We made ambisonic recordings of Singaporean everyday sonic
environments and selected 12 excerpts of 90 seconds duration each,
in 4 categories: city parks, rural parks, eateries and shops/markets. 43
participants rated soundscapes according to the Swedish Soundscape
Quality Protocol (Axelsson et al. 2011) which uses 8 dimensions
related to quality perception. Participants also grouped blobs
representing the stimuli according to a spatial metaphor and
associated a colour to each.
A principal component analysis determined a set of acoustic features
that span a 2-dimensional plane related to latent higher-level features
that are relevant to soundscape perception. We tentatively named
these dimensions Mass and Variability Focus; the first depends on
loudness and spectral shape, the second on amplitude variability
across temporal domains.
A series of repeated-measures ANOVA showed that there is are
patterns of significant correlations between perception ratings and
the derived acoustic features in interaction with personality
measures. Several of the interactions were linked to the personality
trait Openness, and to aural-visual orientation. Implications for future
research are discussed.
I.
BACKGROUND
594
II.
METHOD
A.Soundscape Stimuli
595
name
description
LeqA
76
73
77
69
52
55
69
73
56
58
84
74
B. Participants
596
score
group
Singaporeans (N=32)
Norwegians (N=11)
normative data (N=1813)
RESULTS
1
III.
A.Participants
0
Openness
Conscientiousness
Extraversion
Agreeableness
EmoStability
597
-5
294
56
299
196
488
199
292
0.05
284
380
68
379
320
487528
104 156
348
109
180
295145
95
330
142
124
370
17
250
346
58
501
285
336
353 365
300
141
185
21 345
22 534 53
321
381
6
497 190
355281
306354
280
328
252
108431
277
505
416
107
343
100
18
184 516
11 80
67 136
282
364
103
278 160
163
16
304
290
260
149
288
62
72
246
428421
461
410
420
506
228
146
244422
446 286
494
444
148
31 248
88 28 342
105
274
83
164
73 157
193 139
77
96
189
191 50
154
117
341
125
502
242
137
335
36 311
97 41576
455
325
445
98
19
262 213
329
214
44283
465
384 307 194
326 480
331
153
448
24
204
90
322
115 152 61
43
209
473
317
79
151
256
46
4
40
268 144
334
409 289511
533
131
316
337
273
134
466
332
493
140
138
54
359
368
74169 476
520
1279155
405
216
52
212 349
310
81 113
263
427
440
10
419 247
236
120 170
70
251 26
372239
240 508
327
309
272264
59
82
174
333
367
30
215
385
517 318
523
101
382
253
128
78
192
37
351
187
472
269
340
442
301
32 241
362 226
42
102
439
270 186
114 366
400 261
234
324
235
224433
509
356 221
182
49
229
147
208
477
500
432
220
314
469
237
369
158
271
172
110376 338
408
496
4629
319 133
233 460 33
238
374
118 510
259 181 13
424 211
305
434
470
394
47
232
302
361
223
468
537
475
5 106
129
41
57
205
266
412
452
276
436 130
397
179
257
413
230
219
217
406
92
112
69
34
225
254
206
150
178
7
14315
218
491
267
425 198 453 401
429
464 15
66
202
503 479
132
430 65
371
210
529
162 474
143
227
275
377378
291
161
478
398
166
518
265
29
89
441
513
23
197
514
507
313
489
201
207482
86 122 222
484
2
121
87
94
451
467
363
402
27 99
471 126
396
35
344
84
71
pleasant
12
499
188
0.00
calm
chaotic
annoying
111
423
524
495
463
63
387
347
513417
414
435
monotonous
uneventful
483
303
171
173
486
438
449
418
532
411
165
535
403
426
490
437
-5
75
375
350
25
258
45
52220
388
450
116
119
-0.05
B. Ratings
515
38
512
323
308
360
298
249
64
eventful
exciting
443
383
407
167
357
312
454 358
352
404
60
200
492 391
504
176
456
175 168155
287
297
538
393
177
245
48
390
389
485
243
93
498
519
255 457
339
386
540
527
447
231
0.05
0.00
-0.05
chaotic
exciting
crowd shop
people
annoying
resto
market
hawker caf
pleasant
night oriol
water
bolly
construct
monotonous
calm
uneventful
Figure 5. Plot of SSER soundscapes in SSQP PleasantnessEventfulness, with mean Lab colour.
598
eventful
exciting
21
annoying
monotonous
44
25
24
27
30
31 2 pleasant
19
4028
4
1339
42 16 5
12
35 22
11
2312045 9
18 29 7
10
14
2643
336
632 calm
8 34
3815
17
uneventful 37 41
33
15 33
17
7
32
42 929
16 21
10 6 41
37 24
14
12
35
27 426
23
30
22
23 3113
43
36
58 28
annoying
11
40 44
19
39
18 20
38
1
34
chaotic
25
monotonous
45
exciting
pleasant
calm
uneventful
N10
Soundscape 6 ('night')
Soundscape 8 ('shop')
eventful
chaotic
annoying
monotonous
spectspread -0.95
zeroXrate 0.34
CminusA -0.62
25
2238
24
8 29 28 44
2 26 30 pleasant
27 21
7 12
10
31
1
23
32015 45
16
13
9
11 19 18
45
14
406 39 43
36 35
17 calm
34 32
41
42 37
uneventful 33
25
15
17
exciting
32 41
5
29
7
1637 12
643
4 22927 26
35
11 13 31
134
3
40
14
38
21
23
28 36
8 42 19
18
20
2
annoying
pleasant
39
30
44
chaotic 10 24
33
-0.70
-0.53
0.62
0.45
N10m90
eventdens
0.53
-0.10
-0.67
0.14
0.45 -0.75
-0.58 0.56
-0.22
-0.19 -0.50
tempo
0.19
-0.19
-0.35 0.03
45
monotonous
-0.30
0.67
rolloff
eventful
exciting
0.09
calm
uneventful
0.0
0.2
10
rolloff
-0.6
0.4
2
1
0.2
11
9
2
spectspread
4
8
N10
12
-0.2
-1
1
0
-1
-2
N10m90
CminusA
-2
rolloff
unit description
-3
-0.4
feature
zeroXrate
0.6
zeroXrate
tempo
5 6
eventdens
3
-4
0.2
0.0
-0.2
PC2 (29.2%)
0.4
0.6
-0.4
0.4
PC biplot
of acoustic
features
To find candidates
for computing
acoustic
features, we
-3
-2
-1
0
1
3
investigated-4 MIRtoolbox
(Lartillot
2011)
and2 Genesis
(2009), and a script for ITU loudness from (Nygren 2009).
We identified 24 functions that could3 be meaningful for
measuring soundscapes in terms oftempo
their loudness, spectral
shape, but also rhythm, which should be thought of not so
much in musical terms, but rather
as the rate eventdens
and distribution
12
N10
of events in soundscapes.
With only 12 target soundscapes,
the rule of thumb says that we should
not use more CminusA
than
2
6
8 7
5
dimensions to map them, or else
4 risk over-fitting in the
11 In order to reduce the number, we tried
succeeding analysis.
2
different sets ofN10m90
candidate features and tested them with
principal component analysis. A final selection is listed in
Table 2.
spectspread
1
-2
-1
599
0.0
eventful
chaotic
Soundscape 7 ('resto')
-0.4
Soundscape 5 ('bolly')
col.d2m
0.22
0.22
-0.11
0.07
timespent
-0.06
-0.01
0.03
0.05
-0.02
rater feature
stimulus feature
F(1, 449)
overall
overall
Openness
aural-visual
Mass
Mass
313.8
122.7
0.000 ***
0.000 ***
overall
Openness
Variability Focus
27.7
0.000 ***
less Openness
overall
aural-visual
Variability Focus
17.6
0.000 ***
overall
confusion
Mass
9.1
0.0027 **
less confusion
overall
vigour
Mass
8.4
0.0039 **
low vigour
Pleasantness
Openness
Mass
357.5
0.000 ***
more Openness
Pleasantness
aural-visual
Mass
184.1
0.000 ***
Pleasantness
Openness
Variability Focus
31.1
0.000 ***
less Openness
Pleasantness
aural-visual
Variability Focus
27.6
0.000 ***
Pleasantness
fatigue
Mass
9.1
0.0027 **
more fatigue
Pleasantness
confusion
Mass
6.7
0.0098 **
less confusion
Eventfulness
aural-visual
Mass
190.0
0.000 ***
Eventfulness
Openness
Mass
117.6
0.000 ***
less Openness
Eventfulness
Conscientiousness
Mass
10.4
0.0013 **
high Conscientiousness
Eventfulness
Emotional Stability
Mass
9.8
0.0019 **
position spread
Openness
Mass
9.4
0.0023 **
less Openness
colour spread
Openness
Mass
13.6
0.000 ***
less Openness
time spent
Openness
Variability Focus
12.0
0.000 ***
less Openness
600
0
-1
Eventfulness
-2
aural-visual level
high {0.64!}, 26%, r = 0.68
mid {-0.18!0.64}, 30%, r = 0.68
low {!-0.18}, 44%, r = 0.4
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Mass
0
-1
Pleasantness
-2
Openness level
high {1.00!}, 23%, r = -0.7
mid {-1.00!1.00}, 60%, r = -0.69
low {!-1.00}, 16%, r = -0.59
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Mass
IV.
CONCLUSION
2) Eventfulness
601
APPENDIX
SSER soundscape stimuli and raw response data are
available at http://www.permagnus.net.
REFERENCES
0
-1
timespent
Adams, M., Bruce, N., Davies, W., Cain, R., Carlyle, A., Cusack,
P., Hume, K., Jennings, P. & Plack, C. (2008). Soundwalking
as methodology for understanding soundscapes In
Proceedings of the Institute of Acoustics Spring Conference
2008 Widening Horizons in Acoustics, Reading UK, April
2008, pp 552-558
Andringa, Tjeerd C. (2010). Soundscape and core affect
regulation. Proceedings of Inter-noise 2009, Portugal.
Axelsson, sten (2011). Aesthetic Appreciation Explicated. PhD
Thesis, Stockholm University.
Axelsson, sten, Nilsson, Mats E. & Berglund, Birgitta (2010). A
principal components model of soundscape perception.
Journal of the Acoustical Society of America #128 (5),
November 2010.
Bodin, Lars-Gunnar (2012). Meaning and Meaningfulness in
Electroacoustic Music. Keynote address to Conference of
Electroacoustic Music studies (EMS) 2012, Stockholm.
Bresin, Roberto (2005). What is the color of that performance?.
International Computer Music Conference (ICMC), Barcelona,
2005.
Cain, R., P. Jennings, J. Poxon, A. Scott (2009), Emotional
dimensions of a soundscape. In Proceedings of InterNoise
2009, 23-26th August, Ottawa, Canada.
Cain, R., P. Jennings, M. Adams, N. Bruce, A. Carlyle, P. Cusack,
W. Davies, K. Hume and C. Plack (2008), SOUND-SCAPE:
A framework for characterising positive urban soundscapes,
In Proceedings of Acoustics 08 Euronoise, the European
conference on noise control, Paris France, June 2008, pp
1019-1022
Cain, R. & P. Jennings (2007), Developing best practice for labbased evaluations of urban soundscapes, In Proceedings of
Inter-Noise 2007, Istanbul, August 2007.
Davies, W. and M. Adams, N. Bruce, R. Cain, A. Carlyle, P.
Cusack, K. Hume, P. Jennings, C. Plack (2007), The Positive
Soundscape Project, In Proceedings of the 19th International
Conference on Acoustics, Madrid, September 2007.
Davies, William J. and Adams, Bruce, Marselle, Cain, Jennings,
Poxon, Carlyle, Cusack, Hall, Hume & Plack (2009). The
positive soundscape project: A synthesis of results from many
disciplines. Proceedings of Inter-noise 2009, Canada.
Eerola, Lartillot & Toivainen (2009). Prediction of
multidimensional emotional ratings in music from audio using
multivariate regression models. 10th International Society for
Music Information retrieval Conference (ISMIR 2009).
Friberg, A., Schoonderwaldt, E., & Juslin, P. N. (2007). CUEX: An
algorithm for automatic extraction of expressive tone
parameters in music performance from acoustic signals. Acta
Acustica United with Acustica, 93, 411-420.
Genesis S.A. (2009). Loudness Toolbox, for MatLab. http://
www.genesis.fr (accessed 14 May 2012).
Gosling, Samuel D., Rentfrow, Peter J. & Swann Jr., William B.
(2003). A very brief measure of the Big-Five personality
domains. Journal of Research in Personality 37 (2003) 504
528.
-2
Openness level
high {0.00!}, 56%, r = -0.081
mid {0.00!0.00}, 0%, r = NA
low {!0.00}, 44%, r = -0.22
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Variability focus
602
603