Hacker News new | comments | show | ask | jobs | submit login
I analyzed the chords to 1300 popular songs for patterns. This is what I found. (hooktheory.com)
497 points by davec 1805 days ago | hide | past | web | 183 comments | favorite

Nice post, I'm looking forward for the next one. Meanwhile I'll give my 2 cents.

The main problem in analyzing tonal music is that we mainly listen to relations between chords. For instance, in the following progression in C major, A major functions as a dominant of D (D is the dominant of G and G is the dominant of C):

    C A D G C. 
OTOH, in the following progression the same A is the subdominant of E:

    E A B E. 
This means that if a song modulates or there's a tonicization [1] the same chord will have different tonal functions and we'll listen to it differently. Just counting a chord in a song may not be enough if they have different functions.

The number of repetitions also matters. Tonally, the progressions C | C | C | G | G and C | G | C | G are the same as C | G. Is he eliminating repetitions in the analysis?

About using A major in C; you can use it as a dominant of D (see my 1st example) or as a chromatic mediant [2] in C major. Of course, in modern music you can use anything you want, but these two are the most common uses.

And, naturally, the types of chords used will vary according to the music style.

[1] http://en.wikipedia.org/wiki/Tonicization [2] http://en.wikipedia.org/wiki/Chromatic_mediant

Doesn't he allow for this by normalising all songs to the key of C before doing the analysis?

The grandparent is talking about modulation within a song. Unless they normalized each section of each song to C without saying so, modulation is still an issue. For instance (though I greatly doubt it's on their list) Jethro Tull's "The Whistler" has one key for the verse, a different one for the chorus, and yet another for the whistle solo...

Right, all of the songs were normalized to C, which exactly the point. At Hooktheory.com you can examine all of the songs in relative notation (i.e. I IV V in place of C F G) which makes it easier to compare chord progressions without having to bother with their respective keys.

Reason: It's easier to play these chords on a 6 string guitar, which has been the dominant instrument of choice for pop song composers.

The first C chord on a guitar is easy to hit with no finger twisting required. It's also easy to switch between the first C, Am, and G chord, you can even do it quickly and repeatedly while drunk as you can imagine many pop songs are written. The first F chord requires a little more careful finger placement but still easy to get too. Sure enough you hear this over and over in pop songs, some simple sequence of C F G A chords over and over.

Not surprising that the complex guitar chords that require six pencil-thin rubber fingers and a degree in music theory to know how to play aren't heard as often.

Your thinking is backwards here. It's not that the music is written like this because it is played on a guitar, it is the guitar that was created to make simple music easy to play.

This is the reason why the guitar strings have the tuning they have, and not any of the million other possible combinations.

I think you are misunderstanding the article - All the songs were transposed to the key of C for the purpose of analyzing chords, so don't draw conclusions from the actual chord names presented as the most populat. Instead, what the article is saying is that the I, IV, V chords are the three most common (as one would expect from basic music theory). In the key of C, these are C, F, and G, so that is how it is presented.

Unfortunately your explanation doesn't hold water if you consider other instruments.

C major / A minor are even easier on piano. (C major and A minor are relative keys – they contains the same notes, just with a different note emphasized as the root, or tonic.) They're the two keys that are made up of just playing the white keys on the piano. In contrast, G major / E minor is probably the easiest key to play in on guitar (containing G, C, D, Em and Am), and it comes in second place.

IMHO C major / A minor, C scales, etc. start out easier, but don't necessarily stay that way: sometimes it's nice to have a couple of black notes in there as a reference point. Especially if you aren't looking at the keyboard.

Only easy because they are traditionally taught first. This is because from a music theory standpoint the keys in which those chords are tonic have the least number of sharps and flats. From a memory and hand position standpoint chords that have a black key in the middle (D maj. E maj. etc) are the easiest.

From a memory and hand position standpoint chords that have a black key in the middle (D maj. E maj. etc) are the easiest.

Disagree. You can move around the chords on the white keys without changing your hand shape and everything will sound harmonically related and therefore in key, plus it's easier to count on the white keys. I've been playing music as an adult for 15 years (and had 5 or 6 years of piano lessons as a kid), I know a great deal of music theory and am familiar with a wide variety of scales, modes, and alternate tuning systems, and I still find it easier to move around on the white keys. I happen to particularly like the Phrygian mode, but tend to just play things in E Major when I'm trying out ideas and then transpose afterwards, because as long as I'm playing on the white keys and remember what the root is I literally can't hit a wrong note.

Some people just stick with the original key that lets them play this way; music teachers refer it as 'white note fever.'

Em is probably the very simplest guitar chord to play, only needing 2 fingers beside each other. Yet it doesn't come in at the top of the list. So I don't think your explanation is the dominant force here.

No way. Em7add11 is waaaayyy easier if you use the right voicing.

Yeah, if you went by novice guitar player easiness, the keys of G and D would be at the top, not C.

Also, barre chords enable you to play in any key you want, so it's not like playing in C# is particularly harder than playing in C.

Except that C is a (the?) natural third cord to use with G and D. And then it's a bit tricky to not get something that sounds like a John Denver song.

Don't leave out the fact that these are the "home row" power chords too - even easier to play and sound like you have an idea about what you're doing. Many popular songs (and of course 'garage band' simpleton songs) are unimaginative power chord arrangements.

I think it's simpler than that:

(1) People learn to play guitar by playing popular songs. These popular songs feature a subset of chords in certain combinations that sound good (eg. C->G->Am->F, C->Am-F-G, Am->F->C->G).

(2) People start creating their own songs using the mental maps of "what chords sound good together" from step #1.

(3) Rinse, repeat

My feeling is that it helps that these patterns are easier to play in C, but I think it's really about matching the "what chord comes next" pattern to a particular key.

I'm going to have to disagree with you pretty strongly. F is just about the most difficult basic chord to play. In fact, besides B flat, it's the only one on the graph that requires four fingers (or a barre). Similarly, A minor is much easier to play than C, yet it is far less common.

The popularity of C and Em as keys is undoubtedly a result of what's easy to play on a standard guitar with no capo, but I see no correlation between chord popularity and ease of playing.

Playing an F requires neither 4 fingers or a barre; a chord only requires 3 notes. Playing an F with 4 fingers is playing the root note twice, you can drop the high F (the hard one) and you're still playing a full F chord. The full barre has the root note in the chord 3 times, only one is necessary.

Technically, yes, but a) that generally sounds weak, and b) it increases the difficulty of playing the chord because it means you have to either mute or pass over the first string.

Muting the first string is natural, trying not to mute it and fret it correctly is what makes it hard to play. It's vastly easier to just play the triad on the 2nd, 3rd, and 4th strings with the first 3 fingers; forget the bar. And if it sounds weak, turn up the volume or strum harder. :)

Yeah, turning up the volume or strumming harder isn't exactly going to fill out that frequency spectrum.

Perhaps you missed the smiley. In any case, you can drop the low root and add the high one and it still only takes 3 fingers. I generally leave out the high note and play the middle triad and the bass with the thumb Hendrix style as it feels the most comfortable.

But your main point was that the F was the most difficult to play, when what you really mean is your favorite 4 note version of the F is more difficult to play. A simple F triad, or even the triad with the bass is trivial to play, certainly no harder than any other first position chord except perhaps Em.

> certainly no harder than any other first position chord except perhaps Em

Or, of course, Em11, although you might call that 0th position.

There may be some truth to this, but it also goes the other way.

The modern 6-string guitar became popular in part because it was good for the kinds of music people wanted to play. I, IV, V, and vi (C, F, G, and Am) have a long history after all.

It's an interesting idea, that the evolution of music might be guided by an interplay between people choosing and designing new instruments to suit their tastes, and people developing new musical tastes using whatever existing instruments happen to be available.

Only if you're playing open chords. The shape for every chord of the same type is the same in any key when played as a barre chord.

Very good analysis. I wasn't even going to comment, since my reaction (personal impression) to the article was "too bad there might be 2 of the 1300 songs that have chords that don't sound like crap" - since they all do, there might be a couple of dozen well-orchestrated popular songs over the last several decades. (I should add that, despite this fact, I do avidly listen to popular songs and don't have a problem. I wouldn't try to reverse engineer their chord patterns though...)

At least your response tells me why this is the case. Thanks!

That is a fair critique of the article. The vast majority of the songs that came into the analysis were in fact billboard top 100 so that naturally filters out songs with chords that might sound bad (chords that as a rule are not built off of the major scale the song uses). What we found most interesting, however, was that there are certain patterns that really do show up. The one that was most striking, and that we pointed out in the article, is that the iii chord is almost always followed by IV or vi in pop music. There were similar trends for other some of the other chords that we'll be posting in a later article.

Snnnzx.. wha?

This may best thought of as a lexical analysis of 1300 popular novels. E.G. what is the most popular word following the word "it". The key of a tune 'controls' the chords available, using a typical chord progression. A song in the key of C most typically has the progression C-F-G or I-IV-V in roman numerals signifying 1 for the dominant C, and 4 and 5 for F and G respectively the fourth and fifth notes in the key's scale.

More interesting might be what are the most popular chord progressions. E.G. I-IV-V or II-IV-Im. Which is what I was expecting to click through to.

A million monkeys can write a hit in how many years, now? And BTW "it was a dark and stormy night" don't you know.

Charting chord progressions definitely sounds interesting. Would also like to see that based on musical style -- i.e. "blues" vs. "rock".

Already done (but not per musical style). See http://mugglinworks.com/chordmaps/genmap.htm for a generic Chord Progression Map, and http://mugglinworks.com/chordmaps/chartmaps.htm for specific maps for all twelve keys.

That's an interesting site, very nice lessons. But one would have to actually read the material to understand the chart (grin). But that chart is getting on toward what I expected to see at the subject of this HN post.

For those interested in a far deeper analysis of music structure, as well as exploring the possibility of computer composition and creativity, see the work of David Cope: http://www.amazon.com/Computer-Models-Musical-Creativity-Dav...

I like this analysis:


While I do believe the popular songs follow some pattern, I think the chord progression is only a subset. Someone should look into why Call me Maybe is so catchy. Seriously though

Writing likable, memorable songs is about striking a balance between comfortable and interesting. It always irks me when people go from "this chord progression is in a lot of popular songs" to "therefore all you have to do to write a good song is use this chord progression".

The "comfortable" bit is the easy part. Making it interesting is the trick. There are thousands of songs using this same progression that are awful.

That's my main objection to the original post, and similar articles I've seen. It doesn't really tell you anything usable about how to write a good song. At no point as a developing songwriter does looking at a statistical breakdown of chord progression help you take the next step. "It's got the same chords as X, so it'll definitely be good!" is not something you hear good songwriters say.

Edit: To clarify, I think I'm agreeing with the sentiment of your last 2 sentences, as I understand them.


http://www.youtube.com/watch?v=JdxkVQy7QLM - Rob Paravorian on how so many songs have the same chord sequence as Pachelbel's Canon in D

Relevant: Pachelbel rant - http://www.youtube.com/watch?v=JdxkVQy7QLM

When I was young and learning music theory in piano lessons, one day I realized that literally half the songs on the radio used the chord progression "I V vi IV". This was a huge revelation to me! To my dismay, I couldn't find any evidence on the internet that anyone else had noticed this, until very recently I saw the "Four Chord Song" by Axis of Awesome (http://www.youtube.com/watch?v=5pidokakU4I). I was hoping to see some mention of that in this article, but was disappointed.

Hi fferen, I V iv V is indeed a super popular progression. You can see an example of it here: http://www.hooktheory.com/analysis/view/james-blunt/youre-be.... One cool thing about the way we store music (using relative notation) is that we can compare any two songs to see how similar they are harmonically. For example, there are several "similar songs" shown below the I V vi IV example in the link that all use the same progression in different keys.

All the music nerds I've been hanging out with consider this a pretty obvious observation. So maybe it's a bit weird it hasn't shown up on the internet before.

But like a lot of other people in here have been saying: There's more to a song than the chord progression. We might as well have been talking about rhythm - the vast majority of Western music is in 4/4 time. Also, practically ALL Western music uses the 12-note system. Not pointing this at you, but it's annoying to hear all the people in this thread complain about how pop music sucks because it's all based on the same foundation. Practically all computer programs are in some way based on C. Does this mean that all computer programs suck?

It's called a deceptive cadence -- that's the I V vi part. The IV is warmup for the V - I cadence that ends it all. "Deceptive" because vi is one note away from I, but in minor, not major.

It's a well-known pattern; sorry the internet let you down.

Perhaps my google-fu was weak back then; searching now I see posts about it on several sites. Ha, there's even a Facebook group called "Stop using the I V vi IV chord progression" (http://www.facebook.com/pages/Stop-using-the-I-V-vi-IV-chord...).

You can switch to the sensitive female chord progression: http://sixfouronefive.blogspot.com/

I'm amused by the fact that this person doesn't seem to know that chords per se - and the analysis of them - is practically useless because they can and actually do vary from version to version of the same song. The thing that matters is how the chords are related (modes and progressions).

I don't know -- I still thought it was interesting. The author could have spent all kinds of time trying to figure out if certain chords were functioning as substitute dominants and all of that, but then the article would have been so musically technical so as not to appeal to anyone who doesn't have a fairly high-level music theory background.

Knowledge of rudimentary music theory makes this link upsetting in what it's missing. It looks at data in the wrong way.

I hope that other articles on HN don't fall apart like this if you know a little about the subject.

Exact same thoughts here. Statistics analysis is completely irrelevant when it comes to music (and any form of art of course).

That said, the whole attempt is impressive (from an engineer's perspective)

Directions for future research:

1. Normalize for key: Express chords as I, ii, iii, IV, etc. This will permit analysis of chord progressions/exceptions (see below) across all keys.

2. Detect common constructs, e.g., 8-12-16 bar Blues, and analyze for exceptions, e.g., use of vi instead of I, use of V versus V7 as turnaround, etc. (And more interesting exceptions, e.g., resolution to ii or iii instead of I, etc.)

3. Related to #2, search for popular songs that do NOT follow/use common constructs. Are there are common characteristics across "second rank" popular songs (by which I mean "popular but not quite smashes, or short-duration - novelty - smashes")?

4. How much variation in key and/or chord progression is there for each artist?


A. Not much of a surprise that C/Am is the most popular key: It is the most accessible - the white keys of the piano. It is also very accessible on the guitar, once one learns F...

B. ...but surprising to me that E is so unpopular, being "the natural key" of the guitar (E, A, B/B7 being so easy to learn and so common to the Blues). Analysis of key use by decade - or genre - could prove interesting....

C. Keys these days (the days of equal temperament) are chosen largely on accessibility: Can the soloist hit all of the notes important to the key? Can the accompanist make all the chords important to the song? Once upon a time, prior to equal temperament, keys had sounds and feels of their own, but nowadays, with equal temperament, the progression from unison to octave is by steps of absolutely the same value in each key.

The reason people fall back to the I IV V and VI chords (C F G and Am in the key of C) is that going between them creates a false sense of forward motion in the listener without actually going anywhere. Moving from the IV to V creates tension that can be built up and released by resolving to the I, or increased by going to the VI and resolved to the IV. Any combination of those pretty much sounds "good". Variations add the II (Dm) in place of the V or in between the IV and V, or add the III on the way to the IV. It's really simple, and made even simpler by power chords because you don't even have to move your hand shape to play entire songs on rhythm. Leads can then switch between major scale phrases and pentatonic (aka blues) phrases of the minor of whatever key is being played in, (so Am blues over C) and almost anything they do sounds good to the average listener. In the end, you only have to keep their attention for ~40 seconds between hooks and just crutch on the catchy chorus and you've got a hit. But if you analyze most popular forms of music, the above is at their core anyway. It's just more bare bones in modern pop and rock music.

The chord choices in Garageband are the chords found within the scale of C, plus Bb (even the Bdim - but, to be pedant, it should actually be a B7b5). Having Bb you can modulate (or "change the key") to F without having to switch scales etc.

Also, the fact that he found D, E and A among the results is probably because of modulations. It's VERY common for pop songs to modulate a whole step during some chorus near the end [1]. As mentioned, G, F and C are V, IV and I. If we modulate a whole step, from C to D, the V, IV and I are A, G and D. It would be nice to consider those modulations into the research.

About the key choice, I believe it's irrelevant. It depends a lot on what's your instrument (Bb, Eb is easier on brass instruments), your style (lots of Metal songs in the key of E because E is the lowest note on guitar), your tuning (lots of rock bands downtune their guitars to Eb or D etc), your proficiency, and, most important, the vocalists range.


[1] Otis Reeding - My Girl, Celine Dion - Because You Loved Me (actually lots of songs by her), Monty Python/Eric Idle - Always Look On The Bride Side of Life, Talking Heads Nothing But Flowers (If you search "whole step key change" you'll get a bunch)

To be more of a pedant, the Bdim should be a Bm7b5, not a B7b5. Also, The Temptations, not Otis Redding, sang My Girl.

Ooops, yes, you're right on both!

While I think this sort of analysis is really cool and potentially interesting, there really isn't anything non-obvious in this article, assuming one is familiar with basic music theory. Hopefully this is part one and the more interesting material is being saved for later.

The article was down, but I assume this will be relevant: http://en.wikipedia.org/wiki/Schenkerian_analysis

We spent a lot of time doing this sort of stuff to flesh out harmonic and melodic patterns/meaning of pieces while at music school. To (grossly) simplify, it's essentially a form of reduction analysis, but the final step of the analysis is always I - V - I chord progression (tonic - dominant) with the 3 blind mice melody above (stepwise descending). I never found the final reduction particularly useful as, though he had a point about the prevalence of the tonic dominant relationship, it was over blown. The reduction steps were very useful for stripping away flourishes though, in order to see what was happening at a more base level in a piece (we analyzed a lot of Mahler this way).

Kinda like Map/Reduce in some ways.

He's doing stuff at a way way lower of musical theory (just saying things like "what chords occur the most?") than Schenkerian analysis. He's certainly not looking at the shape of a song as a whole.

Schenkerian analysis is a ton of fun, although you have to remember to just use it as a tool and not take it too seriously.

I have a pessimistic theory on melodies that could be enforced by this study. Melodies are sequence of tones that can be remembered and sung. We have 7 or 12 or 5 tones, about the same number of distinguishable time patterns. Memory for melodies seem powerful, maybe allowing up to 40 elements in the sequence.

By pure combinations it seems the space of melodies is very large. But this space is in fact dramatically shrinked by the very strict relationships imposed on subsequent tones and the result could be that "we have finished exploring the space of interesting melodies", we are deemed to repeat ourselves, musical invention is something of the past.

This is a good start, I'd be more interested if these chord patterns were compared against a database of UNpopular songs to see if the what sort of differences in chord distribution correlated with popular songs (although where you'd find that, I do not know). It's difficult to understand what this really quantifies - all the "2nd chord" distribution suggests to me at this point is that there's a large difference between actually playing music vs a random sequence of chords. It's good that you're recapitulating that at least, but not really a striking observation.

Regardless, I will be keeping tabs on this. Hah, totally didn't intend that pun.

How does this compare to what Music Theory says about chords and chord progressions? Any Music Theory experts/aficionados around? I've unfortunately forgotten most of what I learned in my one class on it.

He doesn't go nearly enough into depth to say anything surprising. That "C/Am" is by far the most common is not surprising at all; C major is the first key any music student learns.

I wish he separated out the major keys from minor keys. It's easy to tell the difference; almost no-one writes pop music in the natural minor (which shares a key signature with the relative major), as it has no dominant fifth. Rather, melodic and harmonic minor keys are used, which have a distinct key signature from major keys (both have the 7th note of the scale raised to accommodate a dominant fifth chord). This should make them easy to detect.

The "Chord Use" chart also does not say much; C, F, and G are the I, IV, and V chords of the key of C major (the Roman numerals indicating their relative position in the scale), which together form the basic pop progression. Much more useful would have been to bin the chords relative to the key in which the song is written, rather than as their absolute pitch.

What I would also like to see is a histogram of chord progressions, e.g. I-IV-V vs. I-IV-ii-V vs. I-vi-VI-V etc. His preview of next week hints at this, but again he should use relative chord names rather than absolute chord names.

The preview graph, "Chords following em", is not very surprising either. Since this is an absolute chart rather than a relative one, we must assume that the data is reflective of the chord's use in the most common key (C). (Em will rarely occur in a song written in Am; rather, it will be altered to an E7 dominant chord.) Of the other chords available in the key of C, C and G both share two out of three notes with Em and thus are unlikely to be used due to lack of motion between them. Bdim is rarely used in C.

This leaves Dm, F, and Am as likely to occur following an Em. The E and B notes of the Em chord are a half-step below the next higher note in the scale, so they will tend toward upward motion, meaning the next chord will likely contain an F or a C. F contains both of these, hence its prominence in the graph. Both Am and Dm contain one of these notes; however Em->Am is an upward (downward) skip of a fourth (fifth), which is a resolving motion – it is likely that the next few chords following these two will be Dm, G, and C, to continue this motion. Em->Dm, being a skip of a second, does not share this property and hence should not be as common.

I was going to write up pretty much the same analysis of the Em->X progression. I think you aptly covered it. Interestingly, though, as you say Em->C sounds relatively static, I bet you see a lot more C->Em than the reverse (which should suffer from the same problem), and I'm hard pressed to say why exactly.

My guess would be that it's a precursor to an Em->F or Em->Am transition; i.e., if you wanted to transition to a minor mode convincingly from C, you'd toss the Em (or even E7) in there before hitting Am. Example: Santeria by Sublime (I-III-vi-V)

I can't speak for pop music, but you can move from Em to C/E to F, which is basically a delayed resolution of the E to F, but the C Major is definitely there.

Regarding "chords following em": Since all of the songs were transposed to the key of C, em == iii, am == vi, and F == IV. Really what the post says is IV and vi follow iii most often in popular songs.

Well, like the article went into briefly, the I, IV and V chords are going to be the main ones used in a song, especially one from a rock/blues tradition. (a minor, d minor and e minor are the 1, 4, and 5 for a song in a minor key.)

C and a-minor, F and d-minor, and G and e-minor are related chords, and can be interchanged in a standard progression for a different "feel". So it makes sense that e-minor would go to F (V to IV transition) or a-minor (V to I) more than any others. It's interesting that C and d-minor are under-represented in that transition.

It's interesting that the possible chords are all simple triads, with no options for extra tones. A C6 with particular notes in the bass could be interpreted as an a-minor, for example. There's a fair amount of ambiguity there.

Everything in this post is consonant with typical music theory.

The most popular chords are the good old I, IV, and V, and the next most popular are their relative minors (C -> am, G -> em, F -> dm).

The example they give of chords following "em" is also pretty much what you'd expect. The most classic resolution for "em" would be "am" (it's dominant -> tonic in the relative minor key).

The more popular resolution here ("em" to "F") is more of a pop music thing, because it's harder to do classical voice leading with it -- classical composers tended to avoid parallel fifths.

Edited to add: Why mention classical rules at all in a post about pop music? Because there's not much new under the sun -- you'd be hard pressed to find any harmony in pop music that Mozart or even Bach wouldn't recognize.

Good points. Many rules in popular music come directly from classical music (after all, it's been around for a while).

However, as you can see in the Hooktheory database, there are many progressions that seem to be unique to popular music.

You mentioned iii->IV, which is a no-no in classical music due to the parallel fifths (more common: V/vi -> IV). But also look at progressions like: C G Bb F, which is extremely common in pop. Here we have a double plagal cadence set up by the dominant that was never used by Bach or Mozart, which is probably best functionalized as: I V IV/IV IV

I would make the argument that there is little in the way of pop music that strays from traditionally accepted chord progressions. Dissonance isn't something that appeals to a broad audience, and you'll be hard pressed to find someone who can point out an interesting choice in progression for a song on the radio.

Sorry in advance for the pedantry. ;) Dissonance is when notes "clash" with each other and in traditional voice leading dissonance always resolves to consonance.

Atonality, on the other hand, is a more recent development in Western music and is defined as the lack of a tonal center. Atonality can be an extremely interesting way to write music. Some of the first atonal music (note how it's not markedly "dissonant") can be found in Debussy's Prélude à l'après-midi d'un faune[1] and the prelude to act I of Wagner's Tristan und Isolde [2].

I don't listen to a huge amount of pop music, but is there anyone out there that is at least experimenting with atonality?

[1]http://www.youtube.com/watch?v=9_7loz-HWUM [2]http://www.youtube.com/watch?v=1Bwl3GYBM3E

Sorry, I should have been more clear. I was listing a few things that wouldn't show up on this list and wouldn't be in pop music. Dissonance and interesting chord progressions were the two I called out specifically.

The chord stuff is all perfectly reasonable (in pop music, that is; in a classical piece in the key of C you would have an Em going to an Am way more than to F, for example, and you'd see way more D chords).

I don't buy that Eb/Cm is the third most common key in pop music, though. There's not a whole lot of pop music in minor keys, and Eb is a weird key to play guitar in.

Eb is a weird key to play guitar in.

Van Halen did it all the time. (Tuning down a 1/2 step is not uncommon among heavier rock/heavy metal bands. I'm not entirely sure why though.)

That's cool, I hadn't heard of that before. Here's a link with some possible explanations:


The explanation that E-flat works better with tunings for brass and horns was interesting.

(Tuning down a 1/2 step is not uncommon among heavier rock/heavy metal bands. I'm not entirely sure why though.)

You answered your own question; it sounds heavier.

Eb in original tuned guitar is very hard. But yeah detuning it 1 semitone is a good trick.

> There's not a whole lot of pop music in minor keys

In my informal survey of the first 6 songs I got on the "pop" channel on pandora, fully a third were in minor keys. (Rumor has it by Adele, and I wanna go by Britney Spears)

I think one of the primary issues with popular music is that it's often ambiguous whether it's in a major or minor mode, even when the song starts on a minor chord. Classically, one would mark the transition to a minor mode by the existence of an authentic cadence: V/vi -> vi, or by the raised scale degree 5, but neither of those clues exist in either song that you mentioned.

In songs like My Heart Will Go On, it appears that the verse tonicizes I, where as the chorus centers around the relative minor. However does it make sense to say it switches modes every time the section changes? Maybe not

Keep in mind that, with a capo, any key becomes trivial for a guitar.

I do hear piano in Eb occasionally, but in my years of playing guitar I can't say I've ever come across it in any song I've played.

> Eb with 3 flats, for instance, is slightly (though not statistically significantly) more common than F with only 1 flat.

Ask any singer: F is the hardest key to sing in. Most people who have to sing in F unaccompanied will inevitably go flat over the course of the song without lots of practice.

I would have plotted the frequency of each chord relative to the key. (e.g., count chords as I/IV/V/ii/etc. instead of C/F/G/Dm) This automatically corrects for the relative popularity of different keys seen earlier in the post.

I'd love to build this data into a neural network and see if I can come up with robosongs. It should be quite good at coming up with "What chord should come next?"

So if you open up a book like 'Music Theory for Complete Idiots' there's a section on "Chord progressions" that essentially has what you'd recognize as a state transition table for "what chords will sound good after this chord?" and I'm pretty sure there's a row that looks like [iii => IV, vi].

Coming up with a computerized vocabulary for the elements of coherent large-scale composition structures would be more of an interesting area to research than individual chord transitions, because the latter is really a solved problem.

A friend of mine has a site that does something similar. It uses music theory to generate real-sounding music for musicians to practice sight-reading. https://sightreadingfactory.com/

That has been done indeed, that state transition diagram: see http://mugglinworks.com/chordmaps/genmap.htm for a generic chord progression map.

I'd love to hear a Markov chain. I don't think it'd be pleasant. But it'd be fun.

Quite a few of these online:

MC trained on classical composers: http://www.youtube.com/watch?v=qOZ2Q-Ls48U

Iannis Xenakis was a major 20th century composer who incorporated MCs into some works: http://www.youtube.com/watch?v=DTzOWKaDrVI

Haven't seen a serious effort with pop/rock though. Lyrics might be the sticking point.

I like seeing this data charted out. For hardcore musicians, it's probably pretty common knowledge, but it's interesting to see it in chart form.

It actually makes we want to try writing some things avoiding the most common progressions to see what comes out.

With regards to other genres and other time periods, I think you'd find pretty similar data with what was "popular" at any given time. Although there have been composers who push the limits, and with some success, our brains seem to be hard wired to react well to the mathematical correlations that are present in the chord parings. On the other hand, perhaps the more exposed we become to varied chord progressions, the more pleasing they would sound.

There is a very interesting RadioLab episode exploring the rage that incited at the premiere of Stravinsky's "The Rite Of Spring". Essentially, scientists are learning how the brain reacts to dissonance in music. The story begins about 32 minutes in.


Probably more interesting than the actual notes would be the figured chords and their progressions.

In other words, "C G a F" isn't materially that different from "G D e C" or "F C d Bb". All three are instances of the same progression: "I V vi IV" ...which happens to be the most hackneyed (or "effective", depending on your point of view) chord progression in popular music over the last 30 years.

If you transform each chord progression into its figured representation then you can pick up more significant trends such as the above, or blues changes (e.g. "I I I I / VI VI I I / V VI I I") and then you can start to discern when they rose to popularity and which ones are falling out of favor.

For example, in the 50s and 60s, I have no doubt "I vi IV V" was more popular than "I V vi IV" but I have no way to prove it currently and would love to find out if I'm right or wrong on that.

Hey Jimmy,

Hooktheory stores all of its chord progressions in relative notation so we have tools to answer questions like the one you've posted here. Although the Hooktheory database is relatively small, we have 20 songs that contain I vi IV V, compared with 100 songs containing I V vi IV. Just by visual inspection, of the songs that contain I V vi IV, none of them were written before 1975, which may not prove, but certainly supports your claim.

He transposed them all to C for exactly that reason. Doing what you suggest yields no additional data.

I normally use nyud.net first in such cases: http://blog.hooktheory.com.nyud.net/2012/06/06/i-analyzed-th...

On a related note (corniest pun of the day), the Australian musical comedy act Axis of Awesome made a song called Four Chords Song, which is basically a 47-song medley all using the same chord progression.


They've said that the chords are used by some of the most well known songs in the world, and their performance demonstrates this nicely, as in this clip:


The above link is the official video; but they explain their reasoning better in this live performance:


Yes, given that C major is the most popular key, it follows that C, F, and G are the most popular chords. That's called a I-IV-V chord progression, and is the basis for most blues, which rock (and subsequently pop) evolved out of.

Oh, and a G chord isn't any easier to play on a guitar than a C# chord.

This was an interesting article, but it's really sad to see that the comments here have devolved into the predictable "all pop music is garbage because they only use 4 chords" argument. It's exactly the same thing as someone looking at a Jackson Pollock and determining that it's garbage because "how much talent does it take to throw a bucket of paint on a canvas?".

The point of music, or any art, is to evoke an emotive response from the audience. To equate the "quality" of art to the technical abilities of its creator is essentially to reduce it to something more akin to juggling knives or spinning plates. Art isn't about virtuosity, it's about emotion.

But hey, if that's your thing, there are a ton of Dream Theater records available on iTunes.

Very interesting! Although, for most common chords chart, I think that instead of simply counting the number of times a chord appears, you should have also considered the duration of that chord in the song.

Didn't Pandora radio did the same analysis for its recommendation engine?

Yeah, I play guitar as just a personal hobby, and there are a lot of times when I'll play an "in between" chord on my way to the next chord I would consider to be in the progression. Many times it's not a traditional "chord" you would find in this list, but just something to bridge a key change or just create a feeling of movement.

I would say that this is most often limited to music that falls outside of pop culture, though. Katy Perry doesn't need to create movement or flow because emotional expression on the backing track is not what her music is about. The Raconteurs might, though. But once you get to that point you have to make the distinction between notes and chords.

One exception to the diatonic norm in pop music is Radiohead's "Just", which uses an octatonic scale: http://www.youtube.com/watch?v=-_qMagfZtv8

(ya, I consider Radiohead "pop music")

I was hoping he had found the "Carlos Santana Secret Chord Progressions". No such luck. Did he have Frank Zappa's "Shut and Play Yer Guitar" in his collection?

Melodies are mildly interesting. It's harmonics that are an endless source of mystery.

This analysis misses the forest for the trees: modern pop music is all about stylistic and sonic innovation. Musical forms tend to evolve along certain parameters while holding others constant. In jazz and classic there has (historically) been a lot of exploration in harmony and melody, but the sonic palette has held steady over long periods of time. Pop music is the opposite.

In the world of rap music, the sonic innovation has been extreme. The meaning of a Public Enemey "song" cannot be assessed through this type of analysis.

I’d be interested to see if they observed patterns or deviations from the norm when comparing between music genres and perhaps even eras / decades.

My sense, as a classical and electric (contemporary rock / blues) guitarist is that you'd observe interesting deviations from the aggregate results described in the study.

Digging a bit I found the following research piece which shares some more thoughts on this topic:


That's a good recommendation. I have a feeling it wouldn't change the results much for pop songs.

I know Pandora has done some analysis like this for their database, but I thought it was limited to things like major or minor tonality, upbeat tempo, etc. and didn't delve as much into the nitty gritty harmony. One reason for this might be that these patterns are so universal (spanning lots of genres), that it might not be too helpful for determining what types of music people like. I could be wrong about this though.

Why not add some more interesting visualizations of your data? Try http://d3js.org. For example http://bost.ocks.org/mike/sankey/ or http://mbostock.github.com/d3/talk/20111116/iris-parallel.ht... might be adapted to communicate which chords follow the others in a more dense fashion.

Flesh Map charts which body parts are most commonly referenced by musical genre:


The app is pretty cool: http://www.hooktheory.com/analysis/view/celine-dion/my-heart... It animates the score for both "instrumental" and "youtube" sound.

Interesting but a melody analysis would be way more insightful.

Pop music is all about the simple melody, in terms of impacting a recognizable pattern on the brain. That is why it's popular, and you see the same melodies repeated over and over, and over, and over.

Could you combine this data to algorithmically create the "most common song ever made"?

The Most Unwanted Song [1] and The Most Wanted Song [2] were a music experiment by artists Komar and Melamid. They survey people to find the most and least favorite musical instruments, genres, lyrical content. Komar and Melamid also created most and least wanted paintings by country.

[1] http://www.wired.com/listening_post/2008/04/a-scientific-at/

[2] http://www.wired.com/listening_post/2008/05/survey-produced/

I'm not crazy about this analysis, but the basic tool (http://www.hooktheory.com/) is pretty awesome. Props for including MIDI and sheet music exports.

Edit: signup gave me an error 500, though.

The broad stats are cool, but I find that the value really comes from going back to the individual data points. For example, I want to hear the "most atypical" songs. Are they any good?

I eagerly await an analysis using roman numerals. Most listeners don't care what particular key the song is in. What are the RELATIONSHIPS involved in the chord progressions?

There's a great Youtube video of three guys playing many pop songs with the same three or four chords. I'm behind a firewall now or I would post the link.

Can we have a Markov chain diagram for the chord transition with the relevant probabilities, pretty please?

This is awesome. Although I find the upside-down-and-backwards guitar chord diagrams rather jarring.

a (vi) over d (ii) comes as a surprise to me. I notice a lot of jam players will substitute II or II7 for ii but I thought this work was based on tabs. I didn't think the relative minor was all that common in pop music.

It's interesting how many comments here imply that more complex is better.

Are you surprised?

btw: You can choose whichever chords you like in Garageband. Click the setup button (looks like a gear). Then flip from Track to Song. Scroll down and add some diminished 9ths or whatever.

Eventually, software will write music just for you. And lyrics too.

I wonder what the RIAA thinks of someone amassing a database of the chord patterns of 1300 popular songs.

As long as the songs are not distributed and the database owner owns the songs, then what's the issue? Generic analysis of chord progressions are not copyrightable, altho tabs for specific songs in print probably are, but I don't think these are specific song tabs. (but if chord progressions were software, they'd probably be patentable... sigh).

It's the Music Publisher's Association (MPA) that cares about tabs and copyright infringement. The relevant wiki link for some case studies: http://en.wikipedia.org/wiki/Tablature#Legal_issues

I find it hard to believe that D major wasn't one of the more common chords to follow and E minor.

Note that he's only searching songs written in C Major. A D Major would be outside the key so naturally it would be quite rare.

The results are exactly as I would expect. F is the most common chord to follow E, because that's the primary way to resolve the dissonance of that note in C major. When E appears in other chords in C Major it will also tend to resolve to F unless there's a good reason not to (cadential 6-4).

A minor is the next most common after F, because that's the next step in a descending circle of 5ths pattern (E, A, D, G, C)

Ah, I missed that the graph was only showing songs in C major.

The graph showing the chords following E minor was created after all songs were transposed to the key of C major.

Why would the II chord follow iii? Can you cite some examples of that?

The only pop/rock use of II that I can think of is as a turn to drive II7 -> V7 -> I.

Exactly. D Major chord is a V/V in C Major, which has only a couple of common uses in popular music according to the Hooktheory database. V/V -> V7 -> I is one common one. Another is a substitution for ii, like: I -> V/V -> IV -> V, see John mayer, Kelly Clarkson, Kenny Chesney etc.

I agree, I'd really be interested to see what sort of pattern he's thinking about.

I missed that the graph was only showing songs in C major.

I tend to agree, but when you think of it, D major isn't used that much in popular music. When you talk about country or acoustic guitar, sure, but not so much in the Billboard charts.

What I'd like to see is a plot of the number of chords used in a song versus time. Possibly even broken down by genre, or correlated to other qualities (sex of singer) et cetera.

If you start at the 1950s, you'll see very simple rock songs; your classic three-chord rock songs. As you hit the 1960s you'll see more complexity; The Beatles, for instance, had more harmonic complexity than what had come before, which continues to be imitated into the 1970s. Then what happens? I don't know, by the 1980s you're looking at a lot of very simple music again, though music is becoming more diverse genre-wise so you're probably getting a larger spread. Then by the present day you have a disturbing trend of one-chord or even no-chord music; apart from rap [which contains no singing but seems to have got simpler even in the backing tracks over the years] we now find that even sung songs are completely lacking in harmony or chord progression. A particularly annoying example I noticed the other day would be that song (dunno who it's by) with the lyrics "We found love in a hopeless place", which seems to have a melody of just four notes.

I could continue this discussion going backwards in time from the 1950s and talking about how the ever-growing harmonic sophistication of art music through Beethoven to Wagner eventually led to a complete breakdown of the idea of harmony in art music which led to music that nobody liked which led to the death of art music and the establishment of rock and roll from square one, but that's another discussion.

If you're interested in learning music theory (as I am currently) and how the the Beatles' music evolved, this series of articles by Alan W. Pollack is fantastic:


They were originally Usenet postings(!) starting in 1989 or thereabouts, but have been revised a bit. Warning: you could lose hours reading through them all.

One thing that comes up many times there that I'd be interested in seeing added to a quantitative analysis like this one is how many times something can "seem" like it might be in one key but ends up really being in another, but I don't know how you would count that.

Wow, I just read through a couple of those and they look fantastic. I'll bookmark that and read it sometime later.

Harmonic sophistication is a...fraught term. There were jazz musicians who developed a style based on 1 chord, but it needn't be harmonically simple. Ever tried riffing off a humming appliance? It's really fun. I used to do it while waiting for my laundry, since the washers all hummed at the same pitch. When it's you and a fixed tone, your harmonic palette starts opening up in wild ways. Microtones creep in, and new decoration schemes.

You also have to be careful because Bach and Purcell didn't think in terms of our modern harmonic structure. They worked in voicing and figured base, which produces similar, but not identical, results. Go back a little farther, and it's all voicing. Palestrina in in some ways closer to Schoenberg than Haydn.

As for the breakdown of harmony in art music, not so much. Serialism got all the academic and theoretical attention, and many mediocre composers with a theoretical bent worked in it, and a few great composers like Stravinsky and Shostakovitch, but it's hardly all the art music that was being composed. Many composers got through their whole careers without any particular use of it, such as Walter Piston. What that has to do with rock and roll is a mystery to me, since it's an offshoot of jazz, which is alive and well and still serving as a wellspring for new music.

Your analysis is overly dismissive of genres that you presumably don't like. Simplistic pop songs aren't a new phenomenon; they were just as present in the 60s as they are now. We just don't listen to them anymore as they haven't stood the test of time. This was abundantly evident last weekend when I went through a stack of about 100 45s from the 60s and early 70s that I inherited. (My father was a radio DJ during the time.) I'd never heard of 90% of the bands, despite already owning hundreds of albums from the period. The same will be true of the music that's popular now in a few decades.

Also, using harmonic progression alone as a measure of musical complexity or richness is misguided. Steve Reich's Violin Phase contains only 5 notes, but was a seminal piece of minimalist music in the 1960s:


Musical richness can come from a lot of places. Hip-hop, as a genre, tends to focus on rhythmic texture and narative content. I'd say, on average, pop hip-hop these days is richer than pop rock. There's also some really great stuff that's come out of the hip-hop world, depending on how far away from pop you're willing to still call something hip-hop, e.g. DJ Spooky:


Just as an example, I love the texture of a lot of stuff from The Roots:


(Note: Several members of The Roots are competent jazz musicians.)

Even Miles Davis's most known albums, Kind of Blue and E.S.P. represented a step towards more simplisitic harmony – "modal jazz" – far simpler than the hard-bop which was at the time prevalent and which Miles had previously played.

The assertation that harmonic complexity precluded the death of art music also seems off-base. Art music was always high-falutin' stuff that mostly rich people, or those wanting to emulate them, listened to. The rise of the relative importance of pop music had more to do with the invention of the phonograph than the increase in harmonic complexity. In fact, many of the first post-romantic composers, Satie, Debussy and Ives, tended towards more simplistic harmonies (arranged in ways that violated the rules of classical and romantic functionalism). It wasn't until the modern period, well after the rise of recorded pop music, that harmonies got particularly wild.

We just don't listen to them anymore as they haven't stood the test of time. This was abundantly evident last weekend when I went through a stack of about 100 45s from the 60s and early 70s [...] I'd never heard of 90% of the bands

There was a Cambrian explosion of music in that period, and virtually no one ever heard of 90%, probably 99%, of the bands. Countless groups sprang up all over the place and pressed records for their local markets. People have devoted careers to tracking down the recordings of that period; one of them, Greg Shaw (whom I knew for a while) had over a million records. He put out a series of influential compilation albums of his favorites that spawned an entire genre (garage rock). Decades later, this stuff is still being unearthed and released. There are entire series of albums devoted to the 1960s proto-punk of Oregon, or Denmark, or Uruguay. It's just amazing how much there is (enough that I'm skeptical of claims that more was recorded in the 80s), and much of it -- tastes vary, of course -- remains amazingly alive and good. A lot of people would disagree that it hasn't stood the test of time; every generation produces new acts in this lineage (e.g. Black Keys), and the underground history of the music continues to be handed down through the fanbase. Its popularity ebbs and flows in a 10-year cycle or so. Right now it's actually on an upswing.

Edit (by way of response to the rest of this thread, not to your comment): it's foolish to identify complexity with good music. Punk/new wave was a reaction against complicated, highly produced music which used an awful lot of chords. The entire career of bands like the Stooges and the Ramones was a self-conscious mining of the opposite aesthetic. Consider the famous 2-note guitar solo in the Buzzcocks' "Boredom": http://www.youtube.com/watch?v=QoYiQ8Qsozk&t=1m15s. The rest of the band was rolling on the studio floor laughing while the guitarist played it because nobody believed it was possible to do a guitar solo that insane, let alone stick with it for that long. It's as far from complicated as you can get but still a great creation. Or think of John Cale's one-note piano drone that runs through the Stooges' "I Wanna Be Your Dog": http://www.youtube.com/watch?v=BJIqnXTqg8I&t=0m22s. Cale came from La Monte Young and the Fluxus movement -- highly self-aware avant garde early 60s stuff that he put into a deliberately primitive incarnation. Not saying everyone should like it (well... ok yes I am :)), but it's as influential and artistic as anything of the last 40 years.

The only general rule in any of this is that you can't predict what form great art will take. The minute somebody thinks they've figured it out (number of chords? please), the muses jump and bless the opposite.

> There was a Cambrian explosion of music in that period

This is a bit of a simplistic explanation. Underlying this its really just that recording, playing and distributing music at reasonable quality became much cheaper and easier, so we rapidly went from a world where music was incredibly localised and diverse but not an attractive career (because nobody made much money) to one where music was heavily internationalised and less diverse, but more attractive as a career (if you could 'make it').

Sorry if you think thats a nitpick, but I really don't see any evidence that the music of the 1960s was an 'explosion' (in diversity? i assume is what you mean) from what had gone before, just more recorded (probably some good stuff, but also a lot of trash - because suddely you could dream of making money from it).

Come to think of it, "Cambrian" isn't the best qualifer since there wasn't an explosion of species (genres), but of records in a few popular genres. Let's just say "explosion".

This didn't happen just because music was "cheaper and easier" (and you call me simplistic! :)). It happened because of the youth culture of the 60s and the asteroidal impact of rock and roll (better science metaphor?), especially the British Invasion of 1963-64. Kids started bands because it was cool. Most weren't expecting to make much money or become stars. They were in it to imitate their heroes and impress girls. This history is well known to fans and students of the period, and it's documented. Fanatical pop archivists have traveled to places like Kansas City and tracked down members of bands who pressed 300 copies of some 45 they recorded in 1966 and interviewed them about it.

By 1969, there was a sharp dropoff, not because the economics changed (it's not as if electric guitars got more expensive) but because the cultural moment had passed. Rock and roll became "rock" and started taking itself seriously. Bands started putting out slicker stuff that, in retrospect, was far less fresh and exciting. Fans of rock and roll talk about those years as the dark ages. The DIY aesthetic kicks back in bigtime with punk, which was a conscious effort to revive the values of the mid-60s (http://www.youtube.com/watch?v=UHPQI4b0ybE).

As I said, this stuff is all well documented. I learned about it from old fanzines and liner notes, so I don't have websites handy, but it ought to be pretty easy to find out about.

"Consider the famous 2-note guitar solo in the Buzzcocks' "Boredom": http://www.youtube.com/watch?v=QoYiQ8Qsozk&t=1m15s. The rest of the band was rolling on the studio floor laughing while the guitarist played it because nobody believed it was possible to do a guitar solo that insane, let alone stick with it for that long. It's as far from complicated as you can get but still a great creation."

If he hadn't spent twenty seconds hammering out E-B-E-B, the ending Bb wouldn't be special at all!

that song (dunno who it's by) with the lyrics "We found love in a hopeless place"

It's by Rihanna, and I can't even understand how it got released. It's like something that got rejected from how-to-make-techno.tumblr.com for being too basic.

It's got a strong hook- something I have come to believe is nine-tenths of a pop single's success.

don't forget a lot of money and promotion. I think that's at least half

Its incredibly repetitive pop drivel. I wonder if teens have a higher tolerance for repetition than adults. They seem the only demographic with any affinity for this type of stuff.

When you're young, you've heard less music. Which means that things that seem derivative and samey to older people don't to teens.

There is also, in general, more tolerance for repetition in dance music than in other genres.

My impression is that older people stop listening to new music, so at some point that's not true anymore..

I listen to a lot of trance/techno music and other similarly repetitive genres (eg http://www.youtube.com/watch?v=jrAxmOvjSYw&feature=relat...). But I like it to have a few different elements to cycle between, and when there's a big musical build I expect it to climax with something other than the exact same thing that came immediately before the build. I have literally made more interesting things than this over morning coffee, and I don't consider myself any great shakes as a musician.

The only explanation I can come up with for the release of this tune is that the quite-talented Rihanna was either incapable or unwilling to go back to the studio to record anything and there was a looming contract deadline imposed by her record publishers, so her management team took some abandoned recording material from a previous session and farmed it out to a momentarily-popular DJ.

Calvin Harris (producer of the song) is hands down the worst DJ I have ever seen in action. No feeling for the audience and horrible noise.

That's surprising, given that the only other Rihanna song I know ("Umbrella") is... well, it's a bit annoying, but it's got significantly more going on musically.

I think there is one thing you are overlooking: IMO, the reason the 1980s had simple music wasn't anything complicated; it was the electric guitar. Power chords are super easy to use, and with an electric guitar they still sound good.

The electric guitar wasn't a enw things in the 1980s. I suspect the return to simplicity was a combination of two things; punk rock musicians' antipathy to the complex arrangements and elaborate production of disco, and the availability of (relatively) cheap and reliable digital synthesizers, which led to a proliferation of electronic 'one man bands.' It's easy to make all the music in a track yourself with a few electronic instruments, but if your primary skill is in the production of said music, then the rhythm, melodies, and harmony are going to be more basic.

I'd consider power-chords as bolstered single notes rather than actual chords. I'm not sure why people are repeating that 80s music was simple. It was a time when synthesizers were becoming more attainable and there was a lot of experimenting on that front. I feel we're going through a similar technical period with computers handling various functions on stage, now.

Hi planetguy, that is an excellent suggestion, and something we've actually already started working on. One of the things about the database our community is building is that it can be used for so many interesting studies. BTW, the Rihanna song you mentioned came on the other day and I heard someone sing "we found Dove in a soapless place". Couldn't help but laugh.

Is this your analysis? I'm far from a novice in Music Theory (only knowledgeable it exists), but thing I'd love to see next would be generating a song through a Markov Chain, or even by applying a genetic algorithm and having viewers vote on their favorite renditions.

I've always wanted to take the lyrics of popular songs and run them through a Markov Chain, just to see what could get produced. Ultimate goal would be to generate a song, the lyrics and then have users upload videos of themselves singing it out.

Yeah, I'm on the Hooktheory team that wrote the article. I love your idea! We're 100% set up to do that kind of thing. It would certainly make an interesting, fun, social set of posts. Today was our first exposure to the world and the feedback / ideas have been amazing. Thanks for the idea. - Chris

I also came here to suggest a Markov chain generator using your data. I know nothing about music but it seems your analysis also needs to include information about timing (A is followed by B within X ms, C within Y ms, etc.) and some sort of clustering information that would allow choruses to be produced (e.g. identify clusters + cluster timing then information about the chords within the clusters).

Even better would be if you guys released your data and allowed other people to come up with creative ways of analyzing and using it ;) I actually have an idea that I think could work quite well and I'd love to build it but lack the data.

This has been crowdsourced once.

Scott Adams, from Dilbert, once postulated that most hit lyrics are nonsense, and asked his readers to post random phrases to create a song:


A German band picked it up and selected the best random phrases and recorded it:



It's not bad actually.

To expand a bit on the 1980s, certainly the average song might have gotten simpler, but there were still some challenging music being written. Most metal and hard rock might have had simpler harmonic structures, but the complexity was going up as solos became more and more challenging (something that continues into the modern day).

I might argue that the regression towards simpler music starting in the 1980s might be due to MTV and increased media coverage of music stars. Starting with MTV and music videos, to be really successful, your looks started to matter. This gave rise to acts who may not have been as musically talented, but were more marketable because they looked better on TV. That's not to say that music acts were never on TV before 1980, which is obviously not true, but MTV did change the game quite a bit and maybe started this trend of towards simpler music with prettier stars.

I'm pretty sure the punk movement inspired the more simple song structures of the 80's more than anything else did; New Wave was a direct offshoot of punk.

A simpler explanation is simply the changing market and distribution. It was really only in the last two hundred years that composers could support themselves by selling their music rather than relying on patronage, with Beethoven being the first to do so. Cheap recordings and broadcasting have changed the game entirely. You might not be able to name a single popular singer of the 19th century, but that's not because they didn't exist; the practice goes back to the troubadours of nearly a thousand years ago.

It's not that art music has had a big fall from grace and lost its grip on the public's heart, paving the way for rock and pop. It never had that grip in the first place. The musical descendants of Bach and Beethoven are alive and kicking; just look at John Williams.

Are you comparing John Williams to Bach and Beethoven? Williams is a very skilled film composer who's great at borrowing ideas from better composers and reshaping them for film. But his music is very unoriginal on its own.

His point was that there is still complex music being composed in the "classical" tradition, outside of pop, regardless of its actual merits or popularity.

Yes and no. It was easy for Bach and Beethoven to be innovative, less had been done back then. The story of classical music from 1800 to 1952, though, is a story of composers forever looking for new ways to innovate. First they did away with all the old rules of structure, and we didn't miss them too much. There's some great music from this period.

Then, they started doing away with all the rules of harmony. The prelude of Wagner's Tristan und Isolde has a dissonant chord which resolves to another dissonant chord, and it totally blew everybody's mind. By 1913 you've got The Rite Of Spring which avoids almost all traditional harmonies, and it's causing riots in Paris but a few years later that kind of thing is old hat. In the 1920s you've got Schoenberg going "Y'know what? Screw tonality. Let's just play all the notes one by one; at this point you've got music which very, very few people out there can actually enjoy. I mean, I have a CD of Schoenberg's violin concerto, and even the blurb on the back of the CD doesn't seem to like it very much.

Then you've got a few decades of increasingly obscure music until my chosen endpoint of 1952, which is the year when Johnny Cage wins a flawless victory in the "look how unmusical I can make my music" championship by publishing 4'33", the piece consisting entirely of silence. This was the signal that the old musical tradition had finally disappeared up its own butthole. (1952 is also the year when Rock Around The Clock was first written.)

Anyway, John Williams is basically an acknowledgement that if you're going to make big orchestral music which people actually like, then all the good ideas were taken by 150 years ago, and any additional innovation away from that point pretty much just makes it worse [or, at least, less enjoyable by the majority of people].

Oh, it's definitely true that many composers reacted and "innovated" away from harmony, and that rock was developed around the same time. But there was a great proliferation in styles in the beginning of the 20th century. During that same period, you have guys like Rachmaninoff, Holst, and Orff -- different points in the musical space, and still popular today. And then you have Ralph Vaughan Williams, who went back to the time before harmony by creating new modal music.

Popular music and art music have coexisted for a long time. There are a lot of trends happening at once in music, and you've hardly made a case that Schoenberg and Cage are strongly connected to the rise of rock. I don't see why your reasoning can't be used to argue that the overwhelming complexity of the Romantics destroyed art music and handed control over to Jazz.

(And, no, John Williams is just an acknowledgement that anyone I call a descendant of Beethoven is going to sound a bit like Beethoven. I can think of several names innovating in the big orchestral space, all of which sound quite a bit different, and most of which you've never heard of.)

Agree with your thesis, but John Williams is the wrong guy for your point. He's a popular composer too, in his own way.

You could point to the Neo-Classicists -- although Prokofiev's dead, of course. Or there's Penderecki; perhaps a hard sell, but deserving.

Definitely. But a major trend in art music relatively recently has been the mix of foreign and other forms with Western styles (and certainly true of the music I'm familiar with). I could hardly call Karl Jenkins or Tan Dun musical descendants of Beethoven. Perhaps I picked a name based on the wrong criteria.

Honestly, I wouldn't mind hearing more of your discussion about art music and the breakdown of harmony.

I second this.

I think you need to clarify some things.

Do you mean 'rock songs' as in Jimmy Preston's Rock the Joint? Or do you mean 'rock songs' as in rhythm and blues/'black music'/music made by black people? If you mean--well, there is no other way of putting this--music that white musicians were playing that was influenced by R&B/gospel/etc., there aren't only simple songs, as can be heard in Preston's music and many other songs that had added arrangements. Even Little Richard.

This makes me think you are referring to only R&B and gospel (which white musicians named 'rock 'n' roll'). If this is the case, this type of music wasn't exactly popular in the way we think of popular music nowadays, not until white musicians 'appropriated' it.

As you can see, there is some incompatibility with what you're saying. 'Popular songs' did not only consist of rock 'n' roll songs, even though rock 'n' roll was considered popular music in the 50s. I mean, Jo Stafford made so-called traditional pop music, and they weren't exactly simple. If you read about late 70s British 'rock', you'd remember that 'rock 'n' roll' was a dirty word associated with the lower classes, drugs, and misfits, which was not necessarily part of popular music. Unless you mean popular music as opposed to 'classical music' (not the 'classical' from the period within 'classical music', mind you).

I don't want to get into a whole Beatles debate, but Beatles took from many styles and genres. So it isn't exactly fair to compare them to typical rock 'n' roll music, which they themselves started playing and were being trained to play initially.

Well, you'd have to go by the top n most popular songs for a given year, because otherwise you'll see an explosion over time since there were simply more albums recorded in 1980 than 1950, or in 2010 than in 1980. 2010 had a lot of shitty simplistic pop songs, but it probably had more prog-metal releases as well. Actually, if you go by all releases, pop music might have a diminishing impact, since it lacks a thriving independent scene.

Incidentally, the Beatles' influence is overstated--they usually popularized the innovations of others rather than innovating themselves. They were influential by virtue of their popularity, but they weren't as original. Pierro Scaruffi overstates his point perhaps, but has written the quintessential polemic against the myth of the Beatles: http://www.scaruffi.com/vol1/beatles.html

Bring on the art music discussion! 1) What ever-growing sophistication? Everything you can do with chords and harmony was done by Bach. The next 150 years were variations on that theme. 2) That Stravinsky and Bartok, et ali, are still seen as "edgy" says more about orchestras' willingness to educate their audiences than it does about Stravinsky, or Bartok.

And 3), rock and roll didn't kill art music, or rise from its ashes. It's a separate thing all together, with its roots in earlier popular music genres. (jazz, bluegrass, the blues...) Rather there was significant cross-fertilization between rock and the not-quite-dead-yet art music sub-genre of minimalism.

What he's really talking about is chord progression (while he doesn't really say that, it's essentially the description of his first few paragraphs). Most university-level music classes discuss common chord progressions. As he notes, the I, IV and V chords are the most common to appear in a song.

It's a great article, but I think he may have done a lot of work to find out something that is fairly common knowledge lol. Still cool though to have the supporting data. (edit: it would be cool to make this an interactive piece of data presentation to help you write songs. Also, the I, IV and V chords are so popular because they naturally make people feel good. It's why they show up so often in 'pop' music. minor chords have a more depressive quality to them)

Also, if you need proof that certain chords show up often in music, just listen to some Nickelback. Here is a fun link (that I THINK works. my speaker only works in the left side ;) ) http://dagobah.net/flash/nickelback.swf

Hi horsehead, I don't know if you made it to the Music Editor part of the site http://www.hooktheory.com/editor but you can use the Music Editor to write songs using I V IV (type 1,5,4 on your keyboard) and such. One of the main goals of Hooktheory is to make more people aware of these basic elements of music theory and help people incorporate it into their musical endeavors.

This is an interesting generic Chord Progression Map: http://mugglinworks.com/chordmaps/genmap.htm

This page contains the specific Chord Progression Maps for all twelve keys: http://mugglinworks.com/chordmaps/chartmaps.htm

Well it isn't hard to figure out chord progressions, modes and key of any song. If you wish to serve 90% of the crowd, who listen to pop music, simply "catch" C Am F G type of chord progressions. For balance, you can expose and iterate with more patterns and lead yourself up to Joe Satriani/Steve Vai's of the world.

Hmm, one could also use a Markov chain on those 1300 chord progressions....

The problem you'd have is a simple Markov chain would have trouble lining up the chord progressions on the larger intervals that chords line up on. You'd end up with a song that sounded sensible in the microscale but just sort of wandered around at weird intervals.

And if you fix this... well... you'd end up with chord progressions essentially indistinguishable from the sampled pop music. But that would be an awfully complex way to obtain results that you could just as easily do by hand. :) In fact many people can just vamp those chord progressions in real time if you don't expect too much creativity.

That's the problem that David Cope had trying to do a Markov-chain-based recreation of Bach, in one of his many experiments in that endeavor; just wanders and doesn't really sound like much, even if all of the n-grams are legit, because it doesn't produce any kind of phrase structure. He ended up with sampling based on hierarchical context instead, iirc: http://artsites.ucsc.edu/faculty/cope/experiments.htm

The site structure is a little strange, MP3 files here http://artsites.ucsc.edu/faculty/cope/mp3page.htm

I did this a while ago to create an infinite music machine. The chains are just based off of one particular song (Bicycle built for two, which also happens to be the song the IBM 704 sang back in the day) but could be made to work with these chords (The site appears down, so I can't access the content).

You can find the source code here : https://github.com/Feni/Infinite-Music-Machine

Peter Langston did some interesting work in 1989 [1] on algorithmic music generation. Many of those live on in Nintendo games today.

[1] http://www.langston.com/Papers/amc.pdf

I thought of this too while I was reading, but I think the results would be pretty bland.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact