Algorithmic composition to help me get started?

It's forgiven :P

This kind of argument is just a bit of a sore point with me, so I tend to overreact a bit.

To me, taking a algorithmic, preset process aproach to composition is a little like trying to make orange juice from lemons.

Music is not meant to be mathematized, or analyzed as such.

OH NOES ALTERNATIVE PATHS TO MUZIC????????!??!?!??!

* Algorithmic composition

* What works and what doesnt, what chords/progressions/notes are used when/where and why.

These rules are extremely general -- so for a general overview of music theory, you could look at any number of websites off of a google search for "music theory." On the other hand, pick up a 500 page book mentioned in the "books" thread.

Neither or these paths will take advantage of your algorithmic background. I'm no mathematician, but what might help is thinking of melody crafting as a markov chain, as in this article: Improvising Jazz With Markov Chains.

The weights to each choice are defined by the genre.

However, the problem with this is that actually modeling this process is an extremely difficult and tedious process. I used this concept in my piece http://www.youngcomposers.com/forum/nanda-vishnu-puredata-16278-3.html#post297290

* Flowcharts

* Pragmatic, step by step approach - a guide i can follow and get audible results.

Again, a quick google search should provide you with this.

I don't really want to get into this whole debate on what music is "supposed to be" and all those prejudices that seem to be flowing around here. But I'd still like to note: I think many people are overmystifying what it means to create art or "express oneself intuitively" in contrast to doing similar things using a computer. In the end, what we write is always a combination of many outer influences, rational thought (sometimes more conscious, sometimes less), maybe some chance. We always compose in ways that have certain "algorithmic" aspects, just that often we have them so absorbed that we're just not realizing the often quite deterministic patterns we tend to follow. Interestingly, I think this actually tends to happen almost more when people think they are writing "totally intuitively" or "out of their soul". Also one should note that many rather "reputable" techniques of the past centuries were very much algorithmic composition. Writing a motet in the strict contrapunctual style of Palestrina is algorithmic to a great degree. So is writing a fugue in the style of Bach. And it's not that significant whether you're doing those calculations in your head or whether you're using a computer. In both cases it's a combination of personal decisions with predefined structures. And I find it just silly if a person who bashes a computer-based composition for not being "personal" enough or anything like that chooses to still make use of sonata forms, violins, cadences, and chromatic scales, ignoring the fact that all of these are at least as much "unpersonal" outer structures as a computer software is. (And before people classify algorithmic composition as the "easy way" or "unartistic", I'd prefer if they'd actually took some time to look at said music and its practices.)

So that's that. Now, to get a bit closer on topic: Basically, I think there are two approaches to your aim. The one you seem to be going for is to base the algorithmic composition of new music on a software that follows a set of music theory rules that are already written down and defined somewhere. This certainly works (even if it doesn't really interest me personally), but you have to realize that all those rules were generally only created to suit the music after it was already composed, and are always just one single perspecticve of describing said music, depending on the personal outlook of the theorists who devised them. If you want to write music that fits these musical models, sure, that's the way to go.

But I think the more consistent approach would be to not start with theories of music devised by other people, but to even create -them- algorithmically. You could make them up in any way you like really, but if your aim is to imitate already existing styles you could just use actual music as your software's input, instead of theoretical rules. Your program would analyse any music that it receives and build a database of generative rules, which it then uses to create new music.

This approach is not too uncommon either. One of the most explored techniques in this respect are Markov chains. A program simply creates tables of probabilities of one specific note following a set of other specific notes and then creates new music based on said tables. Markov chains of high orders can produce rather good results, but of course the load of data also grows huge with high orders. But this is just one very primitive way of analysing music. You could devise totally different ways to analyse music input and draw rules from them.

While i agree on all you've said and i'm interested in algorithmic composition as a research subject(i've red about Markov chains), it's not my goal to write or use a fully automatic software generating the music(at the moment). I might however write a software to aid composition, like a tool for adding simple harmony and such.

I've been misunderstood - my goal is to learn to compose music on my own. Studying algorithms and musical models("music theory rules that are already written down and defined somewhere") would help me understand and familarize myself with general musical concepts and common rules which have been used in composition of various musical pieces. Whereas using Markov chains to compose music doesnt quite improve understanding of the music itself.

Basically i need algorithms, guidelines and rules which could be used by a human to compose a musical piece.

And i'm looking for a good in-depth books/articles on the subject.

(Someone recommended "Composition for Dummies" - it only barely scratches the surface - introduces the building blocks - without actually telling anything about composition itself)

Again, a quick google search should provide you with this.

You're right. I should search around more.

Most resources i've come up with so far introduce you to the building blocks and make a note of common chord progressions and some other basic stuff.

Not so much of hands-on, step-by-step pragmatic stuff i'm looking for.

I know there has to be a lot more in-depth resources out there.

So i'm confused... if you're simply looking to create, then create. It's no secret. You want to stay "safe?" you have 12 possible notes -- take a subset of those 12 notes, put them in order; change timbres and rhythms at will.

You want to mind-filesystemcheck yourself, take all possible frequencies.

Maybe this thread would give you the nodule-style you're lookiong for: http://www.youngcomposers.com/forum/secret-composer-review-20302.html...

OH NOES ALTERNATIVE PATHS TO MUZIC????????!??!?!??!

Well, I never said he couldn't do it that way, that is just the way I feel about it. Hence the "To me" at the beginning of my statement.

Hello fellow aspiring composers and pianists alike!

I'll start off with a brief introduction:

I'm 22 years old. I started playing piano about a year ago coming from a musically secluded environment (i did listen to some music though!)

Although i enjoy playing the piano and all sorts of different music, i also have keen interest in composition and am i still have no idea how and why music works.

My goal (for now) is to be able to write a generic piece of music.

However, i have no idea how to start composing, i can play some random notes and some common chord progressions like I-IV-V-I or i can try to pick out a melody but it doesnt work for me.

I'm a programmer.

I love flowcharts, order and determinism.

I'd like to come across a pragmatic guide to composition.

A step by step guide to writing all sorts of generic (elevator) music.

It has to be clear and concise.

I'm fully aware that there's no full formal mathematical model on what is considered good sounding to our brains and that there's no single method for composition.

But there must be a whole bunch of established guidelines and algorithms out there already. (just take a look at some algorithmic composition software like cgMusic)

I'd like to ask you to recommend me some books/resources/links on the subject to get me started.

Key points:

* Algorithmic composition

* What works and what doesnt, what chords/progressions/notes are used when/where and why.

* Flowcharts

* Pragmatic, step by step approach - a guide i can follow and get audible results.

I hope after trying out and devising several algorithms i'll have better understanding of music and will be able to do this "on my own" taking detours from established algorithms and being able to write music freely as i like.

Band-In-A-Box may be what your looking for. Download the demo and look for the style "Bossa".

As far as music goes, 'to me' it's all about the melodies.

There are things you won't ever be able to pick up with your intellect that you must pick up with your ear. Ear training is the way to go. :-)

But, if you really like the whole intelligent approach, try out music theory. It might hurt the brain a bit.

It's all about singing. You learn the intervals by singing. Singing is the master instrument. It's amazing how everyone can play their own voice, even if they are not a master at any instrument.

If you can sing it, you can make it. :-)

When you are asking for the rules, algorithms, and guidelines to composing music, what you are essentially looking for is music theory. There are all your rules to harmony and shaping melodies or writing contrapuntal masterpieces. It is up to you to put those rules together by experimenting - it's not like buying a model plane and putting it together by the step by step instructions. Let me tell you this - there are no step-by-step instructions to writing music. I say this because the number of different ways you can approach composition of music are virtually endless. No matter what set of rules you follow, it is not THE set of rules because there are thousands of other variations on those rules. These rules are subjective and in fact I don't even like to call them rules, perhaps something more like "expression of self." So you do it YOUR way. If you want a set of algorithms and rules to follow to compose your own music, you are in effect not composing your own music. You are composing somebody else's music for them - who ever made up those rules that you are following. Music theory is the closest you will come to the rules you are looking for and even these rules are meant to be broken anyways.

I think the first thing you should do, is ask yourself what kind of music you would like to compose. Tell us. Is is Vivaldi? Chopin? Beethoven? Debussy? Schoenberg? Howard Shore? Jazz? Mozart? Vaughen Williams? All of these can constitute the loungy background elevator music you are looking for. You can start by picking out exactly what music (by what composer, or what specific movement/time period) appeals to you and either buying scores to their music or find it on IMSLP.org. You can start picking apart how they shape their melodies, what scales they use, what harmonies they use, what chord progressions they use, how they orchestrate, and most importantly, the effect this all conveys. At the same time, try and compose your own piece (start small!) and try and mimick some of these factors. Do it your own way. Just because you have a very mathematical, procedural, algorithmical mind doesn't mean you have an excuse to find a shortcut that the rest of us composers don't get to use. You will have to learn composition just like the rest of us through experimentation and analysation of other composers' works. Trust us, it WILL work. I cannot stress it enough that that is how it has been done since the beginning of time.

By the way, do you improvise much at the piano?

You're right. I should search around more.

Most resources i've come up with so far introduce you to the building blocks and make a note of common chord progressions and some other basic stuff.

Not so much of hands-on, step-by-step pragmatic stuff i'm looking for.

I know there has to be a lot more in-depth resources out there.

I know just a book you want. Its called: Composing A new aproach

Its filled with exercizes etc, that you do, as soon as you learn something new. Very good book, I read it and got some very useful knowledge.

It's forgiven :P

This kind of argument is just a bit of a sore point with me, so I tend to overreact a bit.

understood.

Xenakis's Synaphai was interesting. His percussion stuff is great. :D

Hi JohnD,

I'm nothing closed to an advanced composer. I think I share a bit similar background to you, so may be I can share some experience with you, and hope that you find something useful in it.

I am learning composition myself. My core profession is also software development. And I do like physic and maths.

Back to your original question. It seems to me that you want a "Compositional procedure". I can understand your feeling that when we first started composing it is not clear whether to start. Unfortunately I would agree with the replies in this thread that such a compositional procedure does not really exist. Even if it does, you will most likely want to break it. (because either the procedure doesn't work and people don't use it, or it works but too many people use it and as a result you don't want to write something exactly like everyone else).

My suggestion is to read the score (or transcribe song from recording) of a short piece that you like, and then analyze the various aspect of it (harmony, form, rhythm, etc.). After repeating it on a number of songs this way, you may conclude the commonality of the elements of these songs, and in particular the things that make you like the songs.

Reading music theory is very important too. Otherwise you won't have the language to describe your observation. But most music theory is not compositional procedure.

It's not clear to me what type of music you want to compose. But from your opening post, it seems that you're interested in "generic elevator music", so I assume it is something like smooth jazz and pop songs. One book I would suggest is "Jazz Composition : Theory and Practice" by Ted Pease. I also have "composing music a new approach" that PhantomOFtheOpera suggests. It's a good introduction to get your feet wet, but my experience is that it doesn't have enough meat to guide you write something like an elevator smooth jazz piece.

Lastly, if you're into scientific explanation of harmony. Try this book : "Harmonic Experience: Tonal harmony from its natural origins to its modern expression" by Mathieu.

Well, here an example of what cgmusic could reproduce as music :

cgmusic_test.sib

cgmusic_test2.sib

cgmusic_test3.sib

dunno, it fit in a musical form...still miss the human touch, historical context,... to be considered as Art. Maybe its just a demonstration of what could be a musical form in an abstract way.

Xenakis's Synaphai was interesting. His percussion stuff is great.

Sorry to be off-topic from the thread, but try Keqrops. I prefer it to Synaphai.

I'll just come up with my own way of doing this.

I'm quite certain there are quite a few predetermined methods and rules which in fact could help me a lot. But i can figure it my own way anyway.

Actually recently i've discovered that playing around and rearranging chords and progressions from songs i play, can produce some very good results - music which sounds pretty good and doesnt even necessarily sound close to the original.

Basically i mainly can stay within the framework of song's chords and accompaniment and get some very neat "improvisations" and "derivative compositions". Most importantly, i'm having quite a lot fun with it.

I'll also try to seek a composition teacher or someone versed in it.

Still it's a start!

Thanks, guys!

Here is some music created by different mathematical functions, some of it is pretty interesting. The Sound of Mathematics

Compositional guidelines? Algorithmic what?... Sorry man but if it doesn't come naturally forget about it.

Mentalities like that keep good composers form even putting pen to paper.

For the record:

THERE IS NO ONE WAY TO THINK ABOUT MUSIC. THERE IS NO ONE WAY TO LEARN MUSIC. THERE IS NO ONE WAY TO WRITE MUSIC. THERE IS NO ONE WAY TO ACHIEVE A GOAL IN MUSIC. etc, etc, etc.

Instead of saying "Nuh-uh!" when someone asks about music in a different way than you understand it, maybe it would be better to explore music in that direction.

It is utterly simple to think of music in an algorithmic way. It also requires a near-encyclopedic knowledge of music theory and compositions to think of it effectively in that way. I've sought this kind of thing for computer music applications, but I lack the knowledge to assert the starting situation.

Compositional guidelines? Algorithmic what?... Sorry man but if it doesn't come naturally forget about it.

Thats pretty harsh, not everyone is a musical genius that just has their ability "come naturally".

If you want to work with flowcharts and algorithmic composition, it's good to try a program like MaxMSP, PureData or OpenMusic - it's a different way of composing to working with a sequencer or manuscript but it's good to try different approaches! You can find some algorithmic composition tutorials here

www.algorithmiccomposer.com