Hit or Miss – Can it be predicted?
Written by bluwasp on May 17, 2018
Can a computer predict which songs would become hits?
Researchers analyzed over 500,000 tracks released in the United Kingdom (songs released between 1985 and 2015), in order to understand the ‘dynamics’ of success. They defined this as making it into the top charts). The research team correlated the success with ‘acoustic features’ and used this to explore the predictability of success. (Royal Society Open Science). They used special algorithms to produce the metadata which was then used to predict the past hits and it did so on around seventy five percent of occasions.
Several trends were uncovered.
- They point to a downward trend in ‘happiness’ and ‘brightness’
- A slight upward trend in ‘sadness’.
Interesting to note that the ‘popular’ songs are becoming less ‘male’. Apparently successful songs exhibit their own set of distinct dynamics. In particular, they tend to be ‘happier’, more ‘party-like’, less ‘relaxed’ and more ‘female’ than most.
However, the difference between successful and average songs is not at all straightforward. In the context of some features, successful songs pre-empt the dynamics of all songs, and in others they tend to reflect the past.
Tell that to Public Image LTD… 😉