Monday, June 25, 2012

I analyzed the chords of 1300 popular songs for patterns. This is what I found.

For many people, listening to music elicits such an emotional response that the idea of dredging it for statistics and structure can seem odd or even misguided. But knowing these patterns can give one a deeper more fundamental sense for how music works; for me this makes listening to music a lot more interesting. Of course, if you play an instrument or want to write songs, being aware of these things is obviously of great practical importance.

In this article, we’ll look at the statistics gathered from 1300 choruses, verses, etc. of popular songs to discover the answer to a few basic questions. First we’ll look at the relative popularity of different chords based on the frequency that they appear in the chord progressions of popular music. Then we’ll begin to look at the relationship that different chords have with one another. For example, if a chord is found in a song, what can we say about the probability for what the next chord will be that comes after it?

The Database

To make quantitative statements about music you need to have data; lots of it. Guitar tab websites have tons of information about the chord progressions that songs use, but the quality is not very high. Just as important, the information is not in a format suitable for gathering statistics. So, over the past 2 years we’ve been slowly and painstakingly building up a database of songs taken mainly from the billboard 100 and analyzing them 1 at a time. At the moment the database of songs has over 1300 entries indexed. The genre and where they are taken from is important. This is an analysis of mainly “popular” music, not jazz or classical, so the results are not meant to be treated as universal. If you’re interested, you can check out the database here. The entries contain raw information about the chords and melody, while throwing out information about the arrangement and instrumentation.

We can use the information in the song database to answer all sorts of questions. In this introductory post, I’ll look at a few interesting preliminary results, but we invite you to propose your own questions in the comments at the end of the article.

Let’s get started.

1. Are some chords more commonly used than others?

This seems like such a basic question, but the answer doesn’t actually tell us much because songs are written in different keys. A song written in C# will have lots of C# chords in it, while a song written in G will probably have lots of G’s. That G chords are more popular than C# chords is likely only a reflection of the fact that it’s easier to play on the guitar and piano. So instead of answering this meaningless question, I’ll answer the slightly more interesting one of, what keys are most popular for the songs in the database?



C (and its relative minor, A) are the most common by far. After that there is a general trend favoring key signatures with less sharps and flats but this is not universal. Eb with 3 flats, for instance, is slightly (though not statistically significantly) more common than F with only 1 flat. Bb only has 2 flats but is way at the end of the popularity scale with only 4% of songs using that as the key.

2. What are the most common chords? Part 2

It’s much more interesting to look at songs written in a single common key. That way direct comparisons are possible and more illuminating. We transposed every song in the database to be in the key of C to make them directly comparable. Then we looked at the number of chord progressions that contained a given chord.

Below we’ve plotted the relative frequency that different chords occurred in descending order.



by Hooktheory.com | Read more: