It’s part heart – part math
The Fan Guru makes use of quantitative and qualitative approaches to identify music that is likely to be well received.
The qualitative aspect involves asking a small sample of people who love music to score a song in terms of likability.
The quantitative portion is a bit more involved. But let’s keep it simple. We begin with the underlying assumption that popular songs share similar features by genre. And if we identify a similar distribution in another song, we probably have a song that is likely to be well received by an audience of that genre.
These features include duration, tempo, loudness variation, harmonic simplicity, beat variation, etc.
But these features are only similarly distributed by genre over time. So what constitutes a “winning” distribution is always changing so our algorithm needed to be robust enough to identify and adjust itself to subtle changes across the feature set.
This happens because the appeal of music across ever-changing genres varies significantly amongst groups of people over time. So recent patterns are given more weight. And patterns are identified through modeling called linear regression.
So, the bottom line is, how good is this algorithm? Well, if we were baseball players, we’d be twice as good of a batter as Babe Ruth was during his entire career (.342 lifetime batting average). But our accuracy has been as high as 90%. That means we can reasonably expect to be able to identify a song’s potential to climb the charts for any given genre.
Now that you have a better understanding of how we “guess” the likability of your music song, maybe it’s time you get in touch with us to learn more about how we can get you the right exposure for your music.
We’re looking for good music to play on the radio!
← Back to FAQs