Acoustic and perceptual features of eating-places . PerMagnus - - PowerPoint PPT Presentation

Acoustic and perceptual features of eating-places. PerMagnus Lindborg Nanyang Technological University, Singapore (asst. prof.) KTH Royal Institute of Technology, Stockholm (doct. cand.) permagnus@ntu.edu.sg www.permagnus.net

Monday, July 29, 2013

Soundscape perception

Lindborg: EAT study How much does soundscape quality matter to the experience as a whole? Which sound sources are positive or negative? What are the acoustic differences between places by category e.g. fastfood/ restaurants, cheap/expensive, Western/Asian etc? Soundscapes are important parts of 'servicescapes' (Novak et al. 2010, North & Hargreaves 1996).

Eating-places in Singapore

EAT112: broad, exploratory EAT15: multi-modal ratings EAT40: (priciness)

Monday, July 29, 2013

Perceptual features Lindborg: EAT study Swedish Soundscape Quality Protocol (e.g. Pleasantness- Eventfulness, PLEV) (Axelsson Ö, Berglund G., & Nilsson M.E.) Acoustic features “notice-events” (De Counseel, B.) simulators for acoustic design projects (Hellström B., Lundén, P.) Music emotion research, e.g. Geneva Music Emotion Scales (GEMS, Zentner, Grandjean et al.) Semantic differential (Osgood C. et al. 1957) circumplex models of affect (Russell J. 1980, Thayer) Music (emotion) perception, ecological approach (Friberg A. et al., Vestfäll D., Andean J.)

feature unit Loudness N10 sone Noisiness zeroXrate rate CminusA dBSPL Spectral shape rolloff Hz spectral spread Hz2 N10m90 sone “events” event density Hz-1 tempo BPM

Sound Pressure Level (SPL) Leq A, C - different spectral shape, based on hearing sensitivity curves at different intensity levels (psychoacoustics ear-physiology)

need for richer & more nuanced measures!!!

Monday, July 29, 2013

EAT112: Categorisations 112 eating-places. On-site SPL, audio, annotations, photography (Oct. 2012). Lindborg: EAT study

Monday, July 29, 2013

EAT112: Categorisations 112 eating-places. On-site SPL, audio, annotations, photography (Oct. 2012).

112 eating-places by categories

Bar&Buffet Cafe Fastfood Hawker Restaurant Chinese MixFusion OtherAsian Western

Lindborg: EAT study

Bar&B Cafe Fastfood Hawker Resto sum Chinese 6 3 5 7 21 MixFusion 4 5 5 13 4 31 OtherAsian 2 3 2 4 12 23 Western 1 11 11 14 37 sum 7 25 21 22 37 112

Monday, July 29, 2013

EAT112: Categorisations 112 eating-places. On-site SPL, audio, annotations, photography (Oct. 2012).

112 eating-places by categories

Bar&Buffet Cafe Fastfood Hawker Restaurant Chinese MixFusion OtherAsian Western

Leq by Food Style (broad) in 118 eating-places

Leq Chinese MixFusion OtherAsian Western 68 70 72 74 76 78 80 82 C-weighted A-weighted

Priciness by Food Style in 118 eating-places

Priciness Chinese MixFusion OtherAsian Western

• 0.5

0.0 0.5 1.0

Lindborg: EAT study

Monday, July 29, 2013

EAT112: Swedish Soundscape Quality Protocol (SSQP) Lindborg: EAT study

• 3
• 2
• 1

1 2 3

• 3
• 2
• 1

1 2 3

SSQP 'qualia' for 112 eating-places

Pleasantness Eventfulness

pleasant e x c i t i n g eventful c h a

• t

i c annoying m

• n
• t
• n
• u

s uneventful c a l m Bar&Buffet Cafe Fastfood Hawker Restaurant

Chinese MixFusion OtherAsian Western

Monday, July 29, 2013

EAT112: ANOVA Lindborg: EAT study

• 3
• 2
• 1

1 2 3

• 3
• 2
• 1

1 2 3

SSQP 'qualia' for 112 eating-places

Pleasantness Eventfulness pleasant e x c i t i n g eventful c h a

• t

i c annoying m

• n
• t
• n
• u

s uneventful c a l m Bar&Buffet Cafe Fastfood Hawker Restaurant Chinese MixFusion OtherAsian Western

ANOVA with Pleasantness, Eventfulness, Occupancy, Overall Sonic, and Overall Visual as response variables to three factors: Situation (Indoors or Outdoors), Food Style (4 different), and Design Style (5 different) revealed four significant effects: Occupancy and Situation (F=5.56, p=0.020*, eta2=0.051). Overall Sonic Quality and Design Style (F=2.90, p=0.026*, eta2=0.10). Overall Visual Quality and Design Style (F=5.54, p=0.000***, eta2=0.18). Occupancy and Design Style (F=2.66, p=0.037*, eta2=0.094). … no significant effect on SSQP ‘qualia’

Rated Overall Sonic and Visual Environments compared with Occupancy

rating 2.0 2.5 3.0 3.5 4.0 4.5 5.0

• ccupancy

Bar&Buffet Cafe Fastfood Hawker Restaurant 0.90 0.92 0.94 0.96 Monday, July 29, 2013

Result: Quality ratings correlate negatively with Occupancy

Rated Overall Sonic and Visual Environments compared with Occupancy

rating 2.0 2.5 3.0 3.5 4.0 4.5 5.0

• ccupancy

Bar&Buffet Cafe Fastfood Hawker Restaurant 0.90 0.92 0.94 0.96

Lindborg: EAT study

1 2 3 4 5 0.88 0.90 0.92 0.94 0.96 0.98 1.00 1.02

Correlation between Occupancy and ratings of Sonic and Visual Environment Overall Quality

mean ratings (Likert scale)

• ccupancy

Monday, July 29, 2013

EAT112: Multivariate regression Lindborg: EAT study r2 adj. Priciness SoundMass VariaFocus Pleasantness 0.025

• 0.028
• 0.092

0.216* Eventfulness 0.020 0.098

• 0.060
• 0.219*

Occupancy 0.073 0.26**

• 0.127
• 0.225*

MVR of response features onto independent variables (ß coefficients)

(SSQP Overall Sonic / Visual Quality not used in MVR bcs heteroskedasticity = non-normality of distributions)

Monday, July 29, 2013

EAT15: Multimodal descriptions Studio ratings (N=31) of 15 places from EAT112. Comparing descriptions of different modality: acoustic, semantic (scales and free-form), visual (position and colour). MVR including personality features of raters [3]. Lindborg: EAT study

Monday, July 29, 2013

EAT15: Acoustic features & SSQP ‘qualia’

• 3
• 2
• 1

1 2 3

• 3
• 2
• 1

1 2 3

PCA biplot of acoustic features (EAT15) (rotated 4.95 rad, mirrored around x-axis)

PC1' SoundMass (35.7%) PC2' VariaFocus (31.5%) TWG StbxW CryJad RafMar CoffB StbxE Wishbo LiuSan BeachB SelClb SelRes Fuller Jacks Vanner Tobys

regularity.bx rhythmpeak.bx pulseclar.bx N10.bx X_Zero.crossingrate.bx spectspread.noRMS.bx

Lindborg: EAT study

• 1.0
• 0.5

0.0 0.5 1.0

• 1.0
• 0.5

0.0 0.5 1.0

Biplot of Qualia in circumplex model (95% confidence ellipses)

Pleasantness Eventfulness 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

pleasant e x c i t i n g eventful c h a

• t

i c unpleasant m

• n
• t
• n
• u

s uneventful c a l m

Monday, July 29, 2013

Lindborg: EAT study EAT15: Free word association perceived evoked feelings: sound

Monday, July 29, 2013

EAT15: Free words (rated in VA quadrants, N=3) compared with SSQP ‘qualia’ Lindborg: EAT study

• 1.0
• 0.5

0.0 0.5 1.0

• 1.0
• 0.5

0.0 0.5 1.0

Biplot of Qualia in circumplex model (95% confidence ellipses)

Pleasantness Eventfulness 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

pleasant e x c i t i n g eventful c h a

• t

i c unpleasant m

• n
• t
• n
• u

s uneventful c a l m

Monday, July 29, 2013

EAT15: Visual sorting task compared with SSQP ‘qualia’

• 0.10
• 0.05

0.00 0.05 0.10

• 0.05

0.00 0.05 0.10 0.15

Interface 'white square', mean absolute blob position (50% conf.ellipses)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Lindborg: EAT study

Monday, July 29, 2013

EAT15: Visual sorting task compared with SSQP ‘qualia’

• 0.10
• 0.05

0.00 0.05 0.10

• 0.05

0.00 0.05 0.10 0.15

Interface 'white square', mean absolute blob position (50% conf.ellipses)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Lindborg: EAT study

• 1.0
• 0.5

0.0 0.5 1.0

• 1.0
• 0.5

0.0 0.5 1.0

Biplot of Qualia in circumplex model (95% confidence ellipses)

Pleasantness Eventfulness 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

pleasant e x c i t i n g eventful c h a

• t

i c unpleasant m

• n
• t
• n
• u

s uneventful c a l m

Monday, July 29, 2013

EAT15: Multimodal description are largely congruent Lindborg: EAT study Congruency as MDS similarity (Kendall’s tau) between

• descriptions. (LAB=colour in CIE Lab space)

SMVFa PLEVz LABz PLEVz 0.299 * (28) LABz 0.166 (34) 0.505 *** (34) POSz 0.218 (28) 0.586 *** (28) 0.552 *** (34) PLEVz = SSQP Pleasantness-Eventfulness LAB = colour in CIE Lab space (3 dimensions) POS = vertical-horisontal visual sorting task SMVF = acoustic features (SoundMass-VariaFocus)

• z = values are z-transformed for each participant

Monday, July 29, 2013

EAT15 Lindborg: EAT study

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

pleasant e x c i t i n g eventful c h a

• t

i c unpleasant m

• n
• t
• n
• u

s uneventful c a l m

9
“like” music
(5) birds
(5) 10
“like” voices
(2) fan
(4) 9
“don’t
like” music
(11) background
(3) birds
(4) good
(2) loud
(2) pop
(2) song
(2) 10
“don’t
like” fan
(11) noise
(5) talking
(4) machine
(3) 6
“don’t
like” noise
(7) kitchen
(3) construcCon
(4) loud
(4) 15
“like” insect
(2)

• utdoor
(2)

peaceful/calm
(3) sound
(4) birds/chirping
(2) 15
“don’t
like” crickets
(10) traffic
(4) air
(2) 5
“don’t
like” blending/coffee
machine
(4) kitchen
noise
(2) vacuum
cleaner
(3) 5
“like” Juice/coffee/boiler
machine
(3) [Pasir
Ris
Beach
Bar] [Robertson
Quay] [Seletar
Country
Club] [Coffee
Bean] [Starbucks]

Monday, July 29, 2013

Conclusions: Lindborg: EAT study

difficult
to
idenCfy
acousCc
features
correlaCng
with
perceptual
features non‐verbal
features
(visual,
colour,
spaCal)
funcCon
equally
well
as
semanCc restaurant
soundscapes:

• ccupancy,
return
behaviour
important
response
features
(semi‐conscious)

also:
physiological
responses sound
sources
strongly
valenced:
can
be
miCgated
and
promoted soundscapes
in
general:

Monday, July 29, 2013

References [1] Axelsson, Ö., Nilsson, M. E. & Berglund, B. (2010). “A principal components model of soundscape perception”. J. Acoust. Soc. Am. 128(5), 2836-2346. [2] Lartillot, O. (2011). MIRtoolbox 1.3.2. http://www.mathworks.com/matlabcentral/ fileexchange/24583-mirtoolbox (acc. 24 June 2013). [3] Lindborg PM. (2013, submitted). “Perception of soundscapes correlates with acoustic features and is moderated by personality traits.” [4] Novak, C.C., La Lopa, J., & Novak, R.E. (2010). “Effects of Sound Pressure Levels and Sensitivity to Noise on Mood and Behavioral Intent in a Controlled Fine Dining Restaurant Environment”. J. Culinary Science & Techn. 8:191-218.

www.permagnus.net www.academia.edu/permagnus thank you!

Monday, July 29, 2013