Les algorithmes : entre quotidien et créativité par Vincent FAVRAT et Pierre LEBECQUE, Musimap, LIEGE CREATIVE, 25.06.15

Jeudi 25 juin

Les algorithmes : entre quotidien et créativité

Prof. Samuel NICOLAY et son équipe, ULg – Mathématique

Gregory REICHLING, Citius Engineering

Vincent FAVART, Musimap

Avec le soutien de :

25th%June%2015%

%

LIEGE%CREATIVE%%

%%

%

INTRODUCING%MUSIMAP%%

DISRUPTIVE%SENSITIVE%INTELLIGENCE%TECHNOLOGY%FOR%THE%MUSIC%INDUSTRY%

%%

Vincent%Favrat%–%%COO%

Pierre%Lebecque%–%Head%of%research

©"2015"Musimap"SA"

What:"disrup3ve"music"technology"using"mul3>layered"intelligence"algorithm"""

Concept:% emulates" the" complexity" of" human" brain" func3oning" towards" music" and"decodes"consumers’"personal"tastes"""

Strengths:% bring" personalized" music" recommenda3ons" to" unprecedented" relevance"level"for"enhanced"musical"experiences.""%

Product:% 4" interrelated" APIs" with" music" recommenda3on," music" mapping" solu3ons,"intui3ve"2D>3D"discovery"and"naviga3ons"tools,"social"networks"implementa3on,"digital"shop"and"streaming"configura3ons""

Sales%points:%innova3ve"solu3ons"to"help"content"industry"companies"market"advanced"music" applica3ons" to" beNer" exploit" their" catalogues," increase" the" market" and" their"respec3ve"shares""

The%company

ObjecSve:%become%a%leading%one%stop%shop%of%intelligence%music%services%and%sell%API%or%company%to%key%market%player%

•  Major"shiP"towards"fully"digi3zed"music"industry"(5,9Bn"$)"enabled"by"technological"convergence""

"•  61%"of"internet"users"license"digital"music"services""•  Digital"market"growth:"streaming"breakthrough"(+51,3%"in"2013"="1Bn$)""

""

"

Music%Market%

source:"Ifpi"digital"report"2013"

!  Clear%momentum%with%a%major%ShiV%to%digital%and%streaming%%

Imagine%that%with%30M%tracks!%

Claas%Brieler%(Jazzanova)%–%music%producer%%%%

1st%compeSSve%phase%closed:%"•  similar"catalogue"of"+>"20M"

tracks"•  5%"online"tracks"="90%"online"

sales"/"streams"""

New%compeSSve%phase:%"•  exploit"untapped"95%""•  guide"users"to"consume"more"

music"according"to"their"tastes""

Market%need:%search%engine%as%compeSSve%advantage%in%B2C%

!  Musimap%aims%to%meet%this%challenge!%

StaSsScs

RecommendaSon:%staSsScs%vs.%personalizaSon%

20+%M%tracks% Providers% 2BN%Consumers

Individual%tastes%%PersonnalizaSon%

!  In%culture,%personnalizaSon%is%the%key%to%grow%sales!%%

PotenSal%licensors%"Music%providers%" Music%industry%"

""

Others%%""

Social%Media%%""

Media%groups%" Search%Engines%%

!  AppeSte%proof%already%secured%

%

Olivier%Lebeau%:%head%of%markeSng%%

>"""25"years"of"experience"in"the"music"industry"

>"""Marke3ng"Director"at"EMI"France">"""general"manager"of"Roadrunner"France"

>"""A&R"director"at"Naïve"Records">"""ESLSCA"Business"degree"in"marke3ng%

Vincent%Favrat%:%COO%%

>  Managing"director"of"Just"Tempta3on"(leader"in"film"music"investment"in"con3nental"Europe)"

>  Secretary"General"at"Suisse"Music"Union"

>  Press"manager"at"the"Montreux"Jazz"Fes3val"

>  MBA"in"culture"management"at"the"EAC"(France)"""

%

%Pierre%Lebecque:%head%of%research%%

>  Degree"in"Sociology"(communica3on)"and"Musicology,"Ph.D"on"Musical"Morphing,"Université"de"Louvain"La"Neuve"

>  25"years"experience"as"a"sound"designer"for"Film,"TV"show"producer">  Produc3on"Director,"Canal+"Belgique"/"Nwave"pictures""

Frederic%Notet%:%chief%technical%officer%%

>  TPM"for"Thalys"and"Eurostar"">  Specialist"in"database"and"networking">  Appren3ceship"programming"languages"">  Studied"Music"direc3on"(+"composi3ons"+"theory)"@"Conservatoire"

Royal"de"Musique"de"Bruxelles"

"

The%Team

!  Musimap%counts%on%an%experienced%team%and%30%musicologists%and%30%music%experts%contributed%to%its%technology%

What’s%the%algorithm%about"

!  From%brain%to%machine%

Hard%sciences%meets%soV%sciences:%dealing%with%emoSons,%%relaSons,%influences%%and%contexts

!  Increased%accuracy%for%all%types%of%data,%e.g.%genres%

Alg

orith

m

I kno

w y

ou

Weighted Comparison algorithm

Original track

Track : % Level of importance

- Keywords +- 40%- priority

- Musical styles Influences- potential fuzz

- Other keywords +- 10%- BPM- Rhythmic structure

------------------------------------------------------------------------------------------------------------------------------------

- Main key +- 5 %- Number of musicians +- 2 %- Instruments similarity +- 2 %- Hit-Parade / Best Of Jury / Live +- 0.5 %- Moods / Rhythmic moods +- 35 %

- potential fuzz- Bandwidth +- 2 %- Date and Place of recording +- 0.5 %- Lyrics +- 1 %- Languages +- 1 %- Artists influences +- 1%

AutomaticSimil

CustomSimil

- Tracks i owne (MyMap)- Tracks i like (Favorites, Playlists, Recommandations,…)- Styles i like (Favorites,…)- Artists i like- Blindtests i play- ... Four%chapters%:%

%

1)"Lexicology"

2)"Musicology"

3)"Emo3ons"

4)"Proper3es""

Database%and%items%calibraSon%%

!  EmoSonnal%descripSon%and%lexicodmusicolgy%of%Musimap%%%%%%%%in%size%and%in%quality%

Data% Musimap%

Musicological"parameters" 50""

Complex"Moods"" 500"

Genres"" 390"

Key"words" 1715"

Contextual"analysis" 100"

Items"proper3es"" 10088""

Rela3onships"" 667’727’351"

En33es"(genres,"ar3stes,"albums,"tracks,"releases,"performances,"influences)""

187’287’643""

Algorithm"with"55"weighted"variables"(mul3>layers"cross>influences"matrix""with"advanced"lexicology)""

=%%

Music%brain%with%1B%synapses%(growing%to%5B)%"

with"""

•  2D"&"3D"innova3ve"and"intui3ve"naviga3on"and"discovery"tools%

"

•  Personnalized"music"profiling""

•  Complex"moods"integrated"in"recommanda3ons"

DemonstraSon%

•  Genre%calibraSon%%%•  Track%calibraSon%%%•  Similarity%search%%

•  Synonymic%lexicology%