Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - FuturePulse (FuturePulse: Multimodal Predictive Analytics and Recommendation Services for the Music Industry)

Teaser

The project will help music companies leverage a variety of music data and content, ranging from broadcasters and music streaming data, to sales statistics and streams of music-focused social media interactions and content, through sophisticated analytics and predictive...

Summary

The project will help music companies leverage a variety of music data and content, ranging from broadcasters and music streaming data, to sales statistics and streams of music-focused social media interactions and content, through sophisticated analytics and predictive modelling services to make highly informed business decisions, to better understand their audience and the music trends of the future, and ultimately to make music distribution more effective and profitable. FuturePulse will offer these capabilities over a user-friendly and visual web solution that will enable the immersion of music professionals in the realm of music data and will support them to make highly informed and effective business decisions.

Work performed

Main activities per WP:
WP1: Open innovation, User Requirements and Design
-First version of Music Industry Innovation Report
-First version of FuturePulse requirements
-Draft of second version of Music Industry Innovation Report started

WP2: Music Data Collection, Analysis and Indexing
-Annotated songsets to feed the indexing machine learning algorithms
-Infrastructure set up to automatically collect the specified data for automatic estimation of High-Level Music Content Analysis from audio
-Data are accessible through SaaS to help the development and the scalability of the predictive analysis and recommendations of artists and tracks

WP3: Predictive Analytics and Recommendations
-Popularity estimation and prediction problem explored, a first set of approaches implemented and tested
a. Using social media and web data, experiments carried out with different methods and models of popularity estimation and prediction
b. Multiple sources of data about artist popularity used to estimate and predict the popularity of artists of interest
c. Different predictive models of popularity have been prototyped and tested
-Initial steps in audience profiling and recommendation:
a. Test component for mood analysis from audio
b. Preliminary exploration of inter-country relations with respect to music trends
c. Graph-based models explored as a means of quantifying artist important
-Preparatory steps conducted to support the business-driven music mining and recommendation pilot:
a. Experimental design of the music platform pilot

WP4: Platform Integration and Application Development
-The outcomes of the requirements analysis assessed and aligned with respect to music data collection management, predictive analytics, recommendations, visualisation and user interaction
-Definition of: architecture, technologies, and APIs to be developed and used throughout the project and the definition of an appropriate visualisation engine
-Technical development of the platform: building and setting up the appropriate back-end services and repositories to be used by components owners
-Integration plan has been designed and followed to ensure a smooth and agile technical implementation

WP5: Pilots and Evaluation
-Use Case scenario: define how each Use Case Pilot will work and will be evaluated
-Tools: timeline, technical feedback, workflow, monthly meetings to define a clear workflow for each Use Case Pilot and enable a quick and clear feedback for the technical partners

WP6: Innovation Management, Dissemination and Exploitation
-Communication and dissemination plan designed to reach the widest possible awareness in music and technology industries
-Initial exploitation plan and activities: SWOT analysis and Business Model Canvas
-Setup of communication and dissemination tools
-Dissemination: workshops, social media activity or interviews in online specific media
-Liaison activities with H2020 projects MARCONI and HRadio

WP7: Project Management
-Day-by-day management, effective work and control of deadlines, work plan follow-up, deliverables and milestones follow-up, project advancement, problems to be solved, decision-making processes, motivation and cooperation
-An amendment submitted to include confidential versions of some deliverables
-Setup of External Advisory Board (EAB) and User Panel (UP)
- KPI follow-up and quality assurance for the overall project
-Peer review workflow for deliverables, risk management and contingency plans
-Definition of quality procedures and risk management
-Data Management Plan

WP8: Ethics requirements
-D81 – Protection of Personal Data: report on ethical considerations regarding social media data
-D82 - Humans: procedures and criteria to identify/recruit research participants, information on the informed consent procedures implemented and clarification whether children and/or adults unable to give informed consent are involved

Final results

Progress beyond the state of the art
Innovate on two problems faced by music industry: a) selection and promotion of music and artists, b) better understanding and engagement of music audience
Actual limitations:
-Fragmented music data are largely fragmented
-Long tail artists, albums and songs and it is missing opportunities to capitalize
-Very little understanding of the underlying preference dynamics, music communities and the particularities in the tastes and trends that arise in different geographies and cultures
-Decisions about the future are based on very simplistic assumptions about how popularity evolves
To overcome these limitations, FuturePulse will build upon advances in a multitude of disciplines, including music/audio analysis and retrieval, data mining and machine learning, psychology and social media analysis
The goal within the FuturePulse project is to improve upon current state-of-the-art in the two last factors: a) collect massive amounts of data of various types from various sources; b) develop new content analysis algorithms that allow using contextual as prior
FuturePulse aims to move beyond existing music recommendation systems in two ways: a) Combine contextual factors and group recommendations to provide businesses with music recommendations that suit their audience b) Measure the impact of music recommendations on business performance indicators
FuturePulse aspires to bring to market a cost-effective and simple-to-use solution for studying the effects of music on the behaviour of consumers in the premises of the business
Expected results
The project will result in several high-quality outcomes that will form the basis for the exploitation plan of the project
A robust and extensible multi-source music data ingestion and real-time indexing framework
A multi-modal music popularity prediction engine
An online music community analysis framework and a music recommendation engine
An integrated scalable cloud-based platform offering the full spectrum of FuturePulse services
Three market-driven applications serving the needs of record labels, event organizers and music platforms

Potential impacts
FuturePulse will constitute a significant leap of progress in music consumption and distribution, since it will be the first project to integrate a number of different signals from both online and physical audiences, with the goals of optimizing music selection and distribution, and creating value for a number of actors in the music ecosystem
Music convergence technology developed by FuturePulse has the potential of helping European music companies and professionals to capitalize upon the recent trends and disruptions in the creative music industry
New services to be developed by FuturePulse clearly leverage the convergence of broadband, broadcast and social media

Website & more info

More info: http://www.futurepulse.eu/.