predicting the emotions expressed in music - ida universe · pdf fileusing machine learning to...

30
30/11/2016 1 Cognitive Systems, Technical University of Denmark Using machine learning to decode the emotions expressed in music Jens Madsen Postdoc in sound project Section for Cognitive Systems (CogSys) Department of Applied Mathematics and Computer Science (DTU Compute) Technical University of Denmark Website: www.jensmadsen.com Twitter: @cogniemotion Email: [email protected]

Upload: ngothuan

Post on 22-Mar-2018

218 views

Category:

Documents


1 download

TRANSCRIPT

30/11/20161 Cognitive Systems, Technical University of Denmark

Using machine learning to decode the emotions

expressed in music

Jens Madsen

Postdoc in sound project

Section for Cognitive Systems (CogSys)

Department of Applied Mathematics and

Computer Science (DTU Compute)

Technical University of Denmark

Website: www.jensmadsen.com

Twitter: @cogniemotion

Email: [email protected]

30/11/20162 Cognitive Systems, Technical University of Denmark

Ubiquity of music

30/11/20163 Cognitive Systems, Technical University of Denmark

Spotify (2006) 36+ million songs, (100+ million users)

Apple music (2015) 37 million songs (17 million users)

Deezer (2007), 40 million songs (16+ million users)

TiDAL (2010), 30+ mio tracks (4+ million users)

YouTube (2005), 30+ mio tracks (1+ billion users)

Globally, digital revenues of music account for 45% of total sales in 2016

Denmark, digital revenues of music account for 74.3% in 2014

Music is with us everywhere

30/11/20164 Cognitive Systems, Technical University of Denmark

Major selling points have been

Catalogue

Quality

User Interface

Functionality

Streaming services

30/11/20165 Cognitive Systems, Technical University of Denmark

Design of a music system

SystemUser Interface

30+ mio tracks

30/11/20166 Cognitive Systems, Technical University of Denmark

• Exploration

– Search (artist, title, genre, emotion, …)

– Interfaces (visual, auditory, tactile representations,…)

• Recommendation

– This is what other people are listening to

– Demographic, Collaborative, Content-based, Context-based filtering

• Hybrid

– Similar artist, tracks, …

• Editorial

– Playlists, events, etc.

Celma, Oscar. Music recommendation and discovery: The long tail, long fail, and long play in the digital music space. Springer Science & Business Media, 2010.

Music system interfacing

30/11/20167 Cognitive Systems, Technical University of Denmark

Does recommendation give us what we actually want?

Do people want to explore large music collections?

Design of a music system

30/11/20168 Cognitive Systems, Technical University of Denmark

Why do people listen to music?

30/11/20169 Cognitive Systems, Technical University of Denmark

1. Regulate their emotional state

2. Self-awareness and finding identity

3. Social bonding/relatedness

Thomas Schäfer, Peter Sedlmeier, Christine Städler and David Huron” The psychological functions of music listening” in frontiers in

Psychology 13 august 2013

Juslin, Patrik N., and Petri Laukka. "Expression, perception, and induction of musical emotions: A review and a questionnaire study of

everyday listening." Journal of New Music Research 33.3 (2004): 217-238.

Rentfrow, Peter J; ,The role of music in everyday life: Current directions in the social psychology of music,Social and Personality Psychology

Compass,6,5,402-416,2012,Blackwell Publishing Ltd

Scientific answer to why people listen to music

30/11/201610 Cognitive Systems, Technical University of Denmark

Mechanisms for emotion regulation

Expressed

emotion

Perceived

emotion

Induced

emotionPerformance

/ expression

Audience

Episodicmemory

Emotionalcontagion

Visual imagery

Evaluativeconditioning

Musical expectancy

Brain stem reflexes

30/11/201611 Cognitive Systems, Technical University of Denmark

Audio signal

Annotations (tags, genre, etc.)

User metadata (age, country, etc.)

Lyrics

Building a music system regulating emotions

Episodicmemory

Emotionalcontagion

Visual imagery

Evaluativeconditioning

Musical expectancy

Brain stem

reflexes

Primary mechanisms

of induced emotions

30/11/201612 Cognitive Systems, Technical University of Denmark

Using emotional contagion for emotion regulation

(Mis)matching the emotions expressed in

music with the emotional state of the listener

AudiencePerformance

Expressed

emotion

Emotional

state

30/11/201613 Cognitive Systems, Technical University of Denmark

We need to know what emotions are

expressed in 30+ mio tracks!

Too much data to annotate!

Technical challenge

30/11/201614 Cognitive Systems, Technical University of Denmark

Create mathematical models that can

predict the emotions expressed in music

Overall goal of research

30/11/201615 Cognitive Systems, Technical University of Denmark

Predictive model of emotions expressed in music

Elicitation of emotions

Modelling framework

Audio representation

ModelPredictions

Mo

de

lin

g

fra

me

wo

rk

User

Internal

representation

of emotion

Eli

cit

ati

on

of

em

oti

on

s

User interface

Decision

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

30/11/201616 Cognitive Systems, Technical University of Denmark

ModelPredictions

Mo

de

lin

g

fra

me

wo

rk

Audio representation

30+ mio tracks

User

Internal

representation

of emotion

Eli

cit

ati

on

of

em

oti

on

s

User interface

Decision

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

30/11/201617 Cognitive Systems, Technical University of Denmark

Feature extraction

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

Pitch

Timbre

Beat

Melody

Chords

Harmony

Rhythm

30/11/201618 Cognitive Systems, Technical University of Denmark

Features describing music

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

Pitch/Chroma (100 ms) 12 300

Mel-frequency cepstral coefficients (23 ms) 20 1290

Loudness (23 ms) 24 1400

Beat/tempo (10s) 1 15

Dimensions No. time windows

30/11/201619 Cognitive Systems, Technical University of Denmark

Predictive model of emotions expressed in music

Elicitation of emotions

Modelling framework

Audio representation

User

Model

Core affect

Eli

cit

ati

on

of

em

oti

on

s

Predictions

User interface

Mo

de

lin

g

fra

me

wo

rk

Decision

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

User

Model

Internal

representation

of emotion

Eli

cit

ati

on

of

em

oti

on

s

Predictions

User interface

Mo

de

lin

g

fra

me

wo

rk

Decision

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

30/11/201620 Cognitive Systems, Technical University of Denmark

Elicitation of emotions expressed in music

User

Internal

representation

of emotion

Eli

cit

ati

on

of

em

oti

on

s

User interface

Decision

• Internal representation of emotions

• Define the question to the user

• Define users response format (decision)

• Scale? Rank?

30/11/201621 Cognitive Systems, Technical University of Denmark

Elicitation of emotions expressed in music

User

Internal

representation

of emotion

Eli

cit

ati

on

of

em

oti

on

s

User interface

Decision

30/11/201622 Cognitive Systems, Technical University of Denmark

Internal representation of emotion

Valence and Arousal / Core affect

Valence

Arousal

Active

PositiveNegative

Passive

30/11/201623 Cognitive Systems, Technical University of Denmark

Valence and arousal / Core affect

Valence

Arousal

Active

PositiveNegative

Passive

30/11/201624 Cognitive Systems, Technical University of Denmark

Predictive model of emotions expressed in musicUser

Eli

cit

ati

on

of

em

oti

on

s

User interface

Decision

Va

len

ce

Arousal

30/11/201625 Cognitive Systems, Technical University of Denmark

Response format for quantifying valence and arousal model

Excerpt

A

Excerpt

A

Excerpt

BExcerpt

K

RelativeAbsolute

30/11/201626 Cognitive Systems, Technical University of Denmark

Predictive model of emotions expressed in music

Elicitation of emotions

Modelling framework

Audio representation

ModelPredictions

Mo

de

lin

g

fra

me

wo

rk

Music database

Feature extraction

Feature representation

Au

dio

re

pre

se

nta

tio

n

User

Eli

cit

ati

on

of

em

oti

on

s

User interface

Decision

Va

len

ce

Arousal

30/11/201627 Cognitive Systems, Technical University of Denmark

Modelling framework

𝒳 = 𝑝(𝐗|𝜽𝑛) |𝑛 = 1:𝑁 ∧ 𝐗 ∈ ℝ𝐷𝑥𝑇

𝒴 = 𝑦𝑚|𝑚 = 1:𝑀 ∧ 𝑦 ∈ −1,1

𝜃κ~ℎ𝑎𝑙𝑓𝑠𝑡𝑢𝑑𝑒𝑛𝑡 − 𝑡(ν, η)

k 𝑝(𝐗|𝜽), 𝑝(𝐗|𝜽′) = න𝑝(𝐗|𝜽)𝝆𝑝(𝐗|𝜽′)𝝆d𝑿

𝐟|𝒳, 𝜃κ ~𝒢P 𝟎, k 𝑝(𝐗|𝜽),⋅ 𝜃κ

𝜋𝑚|𝐟, 𝜎 = 𝛷 𝑦𝑚𝑓𝑢𝑚 − 𝑓𝑣𝑚

𝜎2∀𝑚 = 1:𝑀

𝑦𝑚|𝜋𝑚~𝐵𝑒𝑟𝑛𝑜𝑢𝑙𝑙𝑖∓1 𝜋𝑚 ∀𝑚 = 1:𝑀

Discrete choice model

ObservationsUser

Decision

User Interface

Feature Extraction

Experimental Design

Music Database

u

v

Mo

de

l

Feature representation

Au

dio

re

pre

sen

tati

on

Vale

nc

e

Arousal

Likelihood

Decision Making

Latent Function

Model of Core affect

30/11/201628 Cognitive Systems, Technical University of Denmark

Visualizing the latent function (A-V space)

No. Song name

1 311 - T and p combo

2 A-Ha - Living a boys adventure

3 Abba – That’s me

4 ACDC - What do you do for money honey

5 Aaliyah - The one I gave my heart to

6 Aerosmith - Mother popcorn

7 Alanis Morissette - These r the thoughts

8 Alice Cooper – I’m your gun

9 Alice in Chains - Killer is me

10 Aretha Franklin - A change

11 Moby – Everloving

12 Rammstein - Feuer frei

13 Santana - Maria caracoles

14 Stevie Wonder - Another star

15 Tool - Hooker with a pen..

16 Toto - We made it

17 Tricky - Your name

18 U2 - Babyface

19 UB40 - Version girl

20 ZZ top - Hot blue and righteous

J. Madsen, J. B. Nielsen, B. S. Jensen, and J. Larsen,“Modeling expressed emotions in music using pairwise comparisons.” in 9th International Symposium on Computer Music Modeling and Retrieval (CMMR) Music and

Emotions, 2012.

30/11/201629 Cognitive Systems, Technical University of Denmark

• Defined a psychological based approach of designing of music system

• Investigated elicitation methods of emotions expressed in music

• Designed a predictive model of emotions expressed emotion

Conclusion

30/11/201630 Cognitive Systems, Technical University of Denmark

Thank you!

Website: www.jensmadsen.com

Twitter: @cogniemotion

Email: [email protected]