I don’t have the answer, but I’m interested in the question.
Of course, I’m really interested in automated computer composition, but a few things are blocking my progress in that area. 1) The algorithms I’ve come up with are too computationally demanding. Oh the things we humans could do with more computer power. That’s always a problem when you’re trying to do something that’s never been done before with computers, isn’t it? And that’s part of the reason it’s never been done. 2) I lack an understanding of how we humans perceive music. I think most people do. We can’t create a program that writes music like humans if we don’t understand how humans do it. Without that knowledge, we’re basically creating algorithmic and/or recombinatorial music, which can certainly be interesting and sometimes convincing, but it’s not the Holy Grail of the subject (at least, it’s not my Holy Grail).
Anyway, earlier today I was daydreaming of creating a programming language (for fun), and then I thought, hmmm… what if I create a programming language designed to help with the composing process? And I thought, well… that’s just dumb. But I kept thinking, well, how could the composing process be changed? Currently, I just use Overture to click in notes. I think these days there are two main ways to compose music: 1) Write down the notes. Either click them in to a notation program on a piano roll or a blank staff. Or, be old school and use tangible staff paper and a quill pen. 2) Play the music on an instrument. Piano, perhaps. Or even sing it.
Now, some people say “no, I compose in my head!” Oooh, what a genius you must be! I don’t think. All composers compose in their heads. The “composing process” I’m referring to is a matter of getting that music out. You either play it (and perhaps make a sound recording), or you create corresponding graphical symbols (sheet music) to represent how to play it (for either the computer to play, or other humans). What can make composing completely in the head difficult is mainly memory, not lack of intelligence. Writing down or recording the music helps solve this problem. They are processes to aid you in your act of creation while you compose in your head. If you have a good memory and are able to compose a piece completely in your head, don’t look for any praise from me, I really don’t think that’s a very amazing feat.
(On a side note, I think sometimes the composing process is mystified and romanticized to inhumanly heights by people who just aren’t as interested in it. We once had an article in the paper about a local teenager who composed a piece of music for something, and the writer seemed very amazed that a 15 year old could *gasp* write music. Either the writer was just being gracious, or he didn’t realize just how many young composers are out there, and how good they can be. Really, in any art there’s always talk of certain artists being “geniuses” and “prodigies” but, in my opinion, it’s mostly just a romanticizing. Anyone can become “great” with enough practice (it might even be easier to learn when you’re younger, making prodigies even less amazing). “Greatness” is subjective, and fame is an emergent property. People say “we don’t have any Rembrandts today!” or “we don’t have any Mozarts today!” Yes we do, they just haven’t been dead for hundreds of years yet. These artists are put on such romantically high pedestals it seems impossible to compare them to non-famous artists today. But I think the skill level is definitely there. The fame takes time. And you can be “great” (though probably not famous) at any art you’d like… if you’re willing to put in the hours… and it will take some long lonely hours of practice and study. But I do believe that genius is mostly hard work, not a mystical God-given gift given only to a few fortunate fellows (maybe the desire to to do all the required work is… it’s a gift, and a curse… usually when one daydreams of being a genius, one dreams of it coming easily). I might’ve already said all this is some past blog post, but I believe it and it’s a view that not many people seem to share, I think… as far as I can tell. Really I think it’s because people don’t like to think of fame as an emergent property but rather as something that’s destined for objectively “great” people. And that stems from our natural psychological problem of induction, of trying to find cause-and-effect processes where they don’t exist, of noticing patterns and implying improper things from them. So really everyone should just read The Black Swan: The Impact of the Highly Improbable
Where was I?
Ah yes, the composing process…
So, I’m trying to think another system of music representation. Sheet music is designed for a human to read and play the music back. Piano roll view is kind of an “easier to see” version for manipulating notes on a computer, but almost the same thing as regular notation. Both are really just a graph (or a collection of graphs) of time (x-axis) vs frequency (y-axis).
What if we add another axis?
That would be dumb.
But I’d still like to explore the possibilities of designing some other kind of representational system that’s designed for composing instead of for a human to play back. This system might be jarring for composers to use, at least at first. (“At least at first” hahaha… get it?) It would require composers to think in a different way. But that’s the point. Well, really the point is to allow composers to compose faster, and so the point is to experiment and see if there’s way to compose faster if we think about the process in a different way.
I have no specific ideas for this system right now. I’m at work, and I’m just blathering off the top of my mind to help the time go by. Two not-very-specific ideas I have for this system (I’m not sure if they’re any good):
1) Representing changes instead of just frequencies. What if we said something like “up 2, up 3, down 4, up 2” instead of naming notes?
2) Grouping notes. Right now, if you have 6 notes, you have to write all six notes. But what if we group these 6 notes, and then work with that group, and the changes that happen to that group? And then we can go farther and build groups of groups, and look at how those groups are different from one another, and how they’re the same. What kinds of patterns would we find and how could we work with them to compose new music?
I’m not really sure, and none of this may be very innovative anyway, but I am interested in exploring it and getting into more specifics about it.
The goals of the system would be to:
1) Compose faster (i.e. with greater ease). This would in turn allow us to…
2) Explore more possibilities while composing.
And, if possible:
3) Make composing more fun. And thus, attract more people to the act of composing, and help procrastinators and people with composer’s block.
Well, that’s my blather for today. I’ll continue to post my thoughts on this as I have them…
If you read all that, I have two things to say:
1) What’s the matter with you?!
2) Thank you, you are to be commended for your bravery and endurance.
P.S. It was nice to see Michael Giacchino win the Oscar for best score (even though the presenters had no idea how to pronounce his name). I do love his Pixar film music work, he’s doing some of the best film music today, using those things called melodies.
2 Comments
LanthonyS · March 9, 2010 at 6:10 PM
Representing changes is just the naming of intervals, which is a highly technical way of going about it, and for which (as you may know) there are all kinds of confusing and labyrinthian rules–which most amount to a slower way of saying the same thing. I don’t think that road is goin’ much anywhere. At any rate, some musical notation styles do already use it (look up musical notation on Wikipedia for non-Western types), and I’m not sure it’s amazing.
Grouping notes is handled in (and somewhat synonymous with) phrasing, which Chopin worked with to a high degree, as well as many modern piano-specific composers. However, I don’t know how equipped [m]any computerized composing programs are for dealing with them. Certainly they are a very helpful way of dealing with familiar, repetitive, or “mini-movement” passages.
One (not revolutionary) composing tool is the mix of human and computer. Essentially, humans make the base music, or melody and harmony, or something, and the computer algorithmically generates basslines, harmonies, variations, and so forth (though at the moment the only ones I know of that do this sound much more crude than what you’ve been working on). It’s quite fascinating, and while a lot of it can be done by humans anyway, it saves a lot of time and even has an element of randomness that may not otherwise have shown up in their style.
Have you heard of Emily Bear? If not, look her up on Youtube; the first few videos under her name will probably give an adequate picture. Besides her astounding mechanical piano skills, she’s a prodigal composer (which is not to say that her songs are amazing — only a few are fantastic. But for an eight-year-old…) She’s as Mozarty as any composer that (as you alluded to) is actually in the public eye.
Michael Giacchino is fantastic, absolutely great, and I’m glad he won (Hans Zimmer has enough already anyway). Have you heard his work on the Medal of Honor video games? If not, you should; some of it is better than his film score work. (If you want specific game titles, or even specific songs to listen to, I’m happy to give you a list of the best.)
S P Hannifin · March 9, 2010 at 8:00 PM
Thanks for the comment!
For changes, I was thinking on a greater scale than just intervals (mixing it with grouping), but even if we’re just using intervals, how would one compose that way? Off the top of my head, it seems like it would be a chore. In fact, I’d probably begin mentally composing the way I always do and then I’d go back and translate my melody to intervals, which would probably defeat the purpose. Still, it would be interesting to see if any patterns would emerge. Of course, no patterns are going to emerge that we can’t already here in the music, so just using intervals probably wouldn’t help much.
Phrasing may handle grouping, but it’s not what quite what I mean… I’m thinking of, I suppose, collapsing the group into one entity, so that you can’t even see what it’s made up of, at least not until you need to. (Similarly, on a computer, copy + paste can deal with grouping… but it’s still not quite what I’m after.)
Then you build groups among groups, and deal with the changes between those groups…
But I’m still too vague, I will need to try taking a simple piece or melody and translating it, and then see if I can compose backwards from that… if that makes any sense.
Ultimately it’s not going to be *that* innovative; it’s still going to be a system of representing the same music, the music isn’t going to drastically change.
And it may be that the composing process can only become faster and more fun *if* the process is used on a computer. Kinda like what you’re talking about. At least that’s what I had in mind while writing the post… if the process could ease up composing with pencil and paper, I think people would have developed it many years ago.
But I don’t think computers could just “generate” stuff based on human stuff, instead both human and computer would be helping each other along. The computer generating patterns, the human being the guide and director.
This is really not a new idea at all, it’s been the goal of plenty, and there is software out there (like Band in a Box) that already does it.
I’m just trying to think of another way. A similar way, but a different way. I mean, the overall ideas are the same, but I’m trying to think of how it can be made better with a different way of representing and working with the music…
I think it’s hard to judge someone like Emily Bear (or at least it is for me), because, as you say “for an eight-year-old” … which means there’s some level that eight-year-old is much more likely to be at. But there are some big differences between her and other eight-year-olds besides just the skills… her interests, and her opportunities. There’s a line in my screenplay The Melody Box (though I can’t remember if I left it in there or not, I might’ve edited it out) in which one character says something like “I wonder how many people never become composers because they never had pianos” … similar for people who never learned to surf because they didn’t live near the ocean, etc. It’s a property that’s stated a lot in books about how malleable “intelligence” is. I mean, look at the teachers Emily Bear has had. What if everyone had had those teachers, those parents, that house, etc? Obviously there’d still be some differences, because Emily *is* actually interested. How much her surroundings have influenced her interests (or mine, or anyone’s), I have no idea.
So what am I to think of Emily Bear? Who do I compare her to, or should I even be comparing her to anyone anyway?
I *don’t* like the idea that my thoughts about her (or anybody’s) talent should be influenced too much by age, though.
But it is uncommon, especially at age 8… maybe not as uncommon as people think (I wouldn’t call it rare, just look around at the performing competitions they have; there are all kinds of young piano players out there who don’t make it to Oprah, and that’s not because the kids are turning down the opportunity, it’s because the road to that opportunity isn’t direct, you don’t just get there by “being good”), but uncommon enough for the age and talent to be exploited commercially.
I guess.
I think Giacchino’s Medal of Honor music was his first work I heard, though at the time I just though “hmmm, that’s good music” without looking him up. It really wasn’t until I had to buy the soundtrack to The Incredibles because of its awesomeness that I began noticing his name.