Posted by John on July 23rd, 2012
Imagine interviewing one million people worldwide about their music tastes. Then taking that enormous amount of data, opening it up to data scientists, and using the findings to predict whether or not you will like a song you’ve never heard before. The future is now. The EMI Million Interview dataset promises to radically transform analysis and insight in the music industry, improving the understanding of how artists and their fans connect to the benefit of music lovers everywhere.
The big question: can these interviews enable a machine to recommend a new song to a listener based on an answers to questions, demographics, word associations, and their similarity to past interviewees?
“Given the extremely subjective nature of musical appreciation, predicting what kind of music people will like is a much tougher problem than, say, trying to predict someone’s click-through rate for online adverts,” said the chief scientist involved with the project.
The goal is to develop an algorithm that can predict a listener’s level of appreciation for songs and artists, based on the listener’s demographics, word associations, and the past interviews contained in the EMI Million Interview Dataset.
Could this project take most of the “art” out of music writing, performing and producing and inject far more “science.” Will music become manufactured (some may say it already is)? Or will you be able to find more music that suits your individual tastes?