ABC-DJ

The project aims to provide sophisticated ICT systems to support European creative agencies by enabling them to offer branding services and quality products of such high quality that they will be able to successfully compete with existing industry leaders.

The projects’ objective is to select the most appropriate music for any context based on the audio content of songs, not on manually assigned tags but rather based on tags applied by a powerful algorithm that can identify the most efficient features for listener preference forecasting. This prediction model, developed using machine learning procedures that are highly effective, is at the core of the ABC_DJ system.

The algorithm can predict the coherence of the song with the brand or its musical perception with 80.1% precision. Bearing in mind that 100% accuracy can never be achieved because people always have a different reaction to music, 80.1% is clearly an exceptionally high value for the sector.

THE SOLUTION COULD CHANGE THE WAY CREATIVE AGENCIES ACROSS THE WORLD WORK AND THE WAY THAT MUSIC OF LESSER KNOWN ARTISTS IS USED. UP TO NOW AUDIO BRANDING AGENCIES HAVE RELIED SOLELY ON A MIXTURE OF EXPERIENCE, CREATIVITY AND CONSUMER FEEDBACK.

Reactions to music are complex and often subjective. For this reason, choosing the right song to match to a brand and its audience is complex and not error-free. Thanks to ABC_DJ it will be possible to broaden the range of music and to offer European artists and independent record labels new opportunities to monetise their music (for example via online stores, TV ads, etc.).