Robot Composer......

BBC - Radio 3 - Robot Composer ---> You can listen to the computer generated baroque invention here

Spectrum Press New Music

David Cope - Wikipedia, the free encyclopedia

Does someone know about this strange person?

He has written a program which automatically produces chorals, cantatas in the manner of Bach, or even operas in classical style. His productions are satisfactory, although strikingly uninspired.

Maybe this should make everyone seriously reconsider composing in the style of Mozart, Bach, Beethoven or whoever. If a computer can instantly mass produce these mediocre cantatas that we can only create after years and years of study, innovation should remain the only viable course.

Opinions?

I wonder how much input the computer actually has in "composing" something like that.

Sounds like there are a massive amount of rules and guidelines which it basically fills out between.

The idea actually disgusts me :blush:

Edit: was it composed on a given theme? That explains a lot of it.

That saddens me. :blush:

But I find it funny how they put it in the site, the results:

"Track 1: the music was generated by a computer.

Track 2: the music was Bach's Two-Part Invention in G.

Track 3: the music was written by a human."

So, Bach isn't human? :(

No. Bach was German, not human.

No. Bach was German, not human.

ROFL :P

How can a computer compose a song? It's an art! That's just sad.

How can a computer compose a song? It's an art! That's just sad.

You have to look at it this way - the art isn't the music produced by the computer. The art is the program written to produce music from a computer.

...

Indeed; the ability to process all those various factors involved in writing quasi Bach pieces (counterpoint, harmony, voice leading, etc.) is truly amazing! The piece sounds better than a lot of stuff people write now-a-days when they try to write like Bach (I am very much this way).

Maybe this should make everyone seriously reconsider composing in the style of Mozart, Bach, Beethoven or whoever. If a computer can instantly mass produce these mediocre cantatas that we can only create after years and years of study, innovation should remain the only viable course.

Opinions?

Innovation is always the viable course. But that doesn't mean one has to throw away tonal/modal harmony, good voice leading, counterpoint etc.

As someone else said, the "composition" of this music was in the writing of the program. I doubt that it is a fraction as diverse as the human mind composing, since the human mind must/can analyse it's own compositional behaviour and change at will to be "innovative".

And if by "innovative" you mean reverting to non-tonal, without musical sense "noise" - then I'm sure it would be *much* easier to write a program to do that.

Can this program create a good Mozartish piano concerto? A Beethovian Sonata? I doubt it. And even if it could - could it create one without using the same musical tools in the same places so much as to ultimately bore the listener with monotany of expectation?

No way. The only way it could is if the program writer had more compositional knowledge and imagination than these titans.

You have to look at it this way - the art isn't the music produced by the computer. The art is the program written to produce music from a computer.

...

*sigh*

I suppose. :cool:

STILL, that proves my point that Bach is indeed very mechanical. That's why I don't like the majority of his works. They sound cold to me. (Of course I'm not referring to his wonderful masses or his famous tocatta and fugue in D minor...)

I prefer the warmth of romantism! :cool:

STILL, that proves my point that Bach is indeed very mechanical. That's why I don't like the majority of his works. They sound cold to me. (Of course I'm not referring to his wonderful masses or his famous tocatta and fugue in D minor...)

I prefer the warmth of romantism! :cool:

Hahahahaha you suck :) there's as much passion in Bach as in Berlioz or even more, but you can't figure it out.

Yes, he thinks passion is only swooning dynamics.

No. I think passion feels different for each person. Many of Bach's works are pure beauty to me, as others are just mechanical. But I sense no passion in any of them. Like atonal music is just noise to me, and most of Mahler's and Wagner's works give me the urge to shoot the conductor. It's my interpretation. Everyone has his own.

I wonder how much input the computer actually has in "composing" something like that.

Sounds like there are a massive amount of rules and guidelines which it basically fills out between.

The idea actually disgusts me :P

Edit: was it composed on a given theme? That explains a lot of it.

So far as I know, there's no direct human input involved. The user only loads the scores of whichever composer he expects the imitation to be composed, and the machine then begins to churn out masterpieces at a frantic pace, only limited by the user's directions (i.e. happy, sad, length.....)

STILL, that proves my point that Bach is indeed very mechanical. That's why I don't like the majority of his works. They sound cold to me. (Of course I'm not referring to his wonderful masses or his famous tocatta and fugue in D minor...)

I prefer the warmth of romantism! ;)

Toccata and Fugue in D minor may not after all be a work of Bach. There's no surviving original manuscript, and as it abounds in parallel fifths, it's more than likely that it was attributed to him once its real creator faded into oblivion...

Innovation is always the viable course. But that doesn't mean one has to throw away tonal/modal harmony, good voice leading, counterpoint etc.

As someone else said, the "composition" of this music was in the writing of the program. I doubt that it is a fraction as diverse as the human mind composing, since the human mind must/can analyse it's own compositional behaviour and change at will to be "innovative".

And if by "innovative" you mean reverting to non-tonal, without musical sense "noise" - then I'm sure it would be *much* easier to write a program to do that.

Can this program create a good Mozartish piano concerto? A Beethovian Sonata? I doubt it. And even if it could - could it create one without using the same musical tools in the same places so much as to ultimately bore the listener with monotany of expectation?

No way. The only way it could is if the program writer had more compositional knowledge and imagination than these titans.

Yes, it can create a good Mozartish concerto.

I never suggested that one should throw those concepts away...:)

And I don't mean "atonal" by "innovative". Atonalism is the tradition of the twentieth century - it's pretty much clich

From what I know about these kind of programs is that most of the programming is based on Markov Chains based on the composers previous works.

A Markov Chain is a system where probabilities are assigned to different states, so say using a crude example based on Palestrina:

First Note C:

Second Note: D = 30%, B = 30%, E = 20%, A = 20%

And so forth, so you have a 12x12 matrix with different probabilities assigned to each entry, and then the piece is 'calculated' as a chain moving from one note to the next.

It's something I would like to get into. Personally, I think the idea of trying to recreate a Bach-style piece is intriguing and interesting from a mechanical point of view, but kind of useless from an artistic perspective. However, I think that the idea of Markov Chains and Stochastic Systems can be used to create diffreent ideas. I know Xenakis began to look into that, but as I said, its only something I'm just looking into...

Hi..

Western tradition is, as you know, algorithmic not stochastic. And we call the algorithm counterpoint.

I'm pretty sure that using Markov Chains to analyse note by note the development of a bach chorale would hardly ever produce satisfactory results, and even if it did, it would do so through a massive waste of resources. Yes, stochastics and markov chains are used to create music, but that music is fundamentally different in conception from that by which, say Mozart, created his music.

I hope that I haven't misunderstood you...

In my opinion, the simple fact is that composing by hand at this age, while remaining loyal to those principles established in the past, like the avoidance of parallel perfects, is a massive exercise in futility and a pitiful waste of time and resources not because there's anything wrong with those rules, but because a computer is literally a million times better equipped than a human being to deal with the arising mechanical problems of the process, which may well comprise up to ninety-five percent of the burden of composing a piece. Even a work, which succeeds in perfectly following all the rules and principles of the Western tradition, is by no means anything more than a similarly crafted, mass produced piece by a computer, and what is more, we have to understand that we are pretty much at the infancy of the era of artificial intelligence.

Even Gary Casparov, with all his awesome intelligence, was unable to consistently defeat a computer when faced with the prospect of having to face it within the limits of a predefined, well-formed algorithm.

And ultimately, the process of creation within a game of chess, that is, the improvisation of new methods to counter problems arising during the progress of computation, is in no way different from the simple process of creating music. Both are very highly structured, and follow very rigid rules, if we will take the example of Bach or Mozart for instance.

With the advent of the computing age, it's an undeniable fact that the process of composing music is or will soon be reduced to the mere creation of algorithms, as is the case with many other pursuits. Thus we should never allow ourselves to degenerate to the level of a calculator, by confining ourselves to the limits of an idiom, or in other words, a computational algorithm, be it baroque, classical, renaissance or whatever.

And this can all be summarised in a simple sentence: Innovation is the only respectable, worthy path for the composer of our day, incomparably more so than in the past.

To make my point...here're two samples from a Bach chorale composed by David Cope's program. I should as well add that he offers 5000 of these at his website for free. Pretty much the work of a lifetime for a human composer, and they are quite good. Although, as always with computers, there's nothing that prevents the algorithm behind them from being improved.

bach-1.mid

bach-2.mid

bach-3.mid

Hey Neno

Google for Markov Chains and Composing Software and you might find a few things. I know there is a Wiki article about it... and I have definently heard the Palestrina example before.

Western music isn't exactly algorthmic. If it was then the system would be much simpler, and a simple program "IF this THEN that" could be applied. Stochastic systems and algorthims arn't really mutually exclusive.

I'm not sure about the Cope program, as I havn't seen it and I'm not a computer programmer, but I do know that similar programs used to recreate music by Palestrina and even Rachmaninoff to similar effect were based on stochastic matrices.

Maybe this should make everyone seriously reconsider composing in the style of Mozart, Bach, Beethoven or whoever. If a computer can instantly mass produce these mediocre cantatas that we can only create after years and years of study, innovation should remain the only viable course.

I listened to some of the output, and it is pretty good. Actually, is this program available for download? Anyway, as cool as this program is, and although the computer can come up with the harmony, it cannot come up with a decent melody. If it can come up with any structure, it would be severely limited to the programmer's choice. A computer will never match the creativity of a human being in this sense, because a computer can't think creatively. For example, the output of the midi's may sound good, but they are seriously lacking and have no character. Someone like Bach is needed, in order to create a melody against these harmonizations. So in effect, Bach, Beethoven and Mozart are still unique and innovative, since they construct melodies. This program would NEVER be able to produce Beethoven's 5th, or Mozart's symphony #40 in G minor. Never.

In conclusion, harmony can be constructed by algorithms, however melodies, character and free-form cannot. All I see this program being is nothing more than a simple harmonic aid.

David Cope's program can undoubtedly compose melodies, and these examples are harmonic sketches only because they are chorales - one wouldn't expect flashy meloies in a piece of devotional music. At his website he offers samples of his other scores, in the style of anyone from Mozart to Bartok, from operas to concertos, for a fee, but the fact is that a lot of people have already listened to his music, and most admit to being awed by the idea of their feelings being manipulated so powerfully and uniquely by a computer.

What is the difference between having electrodes inside your brain, and being made happy or sad while listening to a piece composed by a computer? It's just that the physical presence of the metal is absent, otherwise there's no difference at all, at least in my opinion.

Moving on to the subject of melodies, I will ask if Bach has melodies at least in his fugues? His music is mostly based on harmonic elaboration through symmetry, permutations and group theory, rather than any kind of divinely inspired Grace. He was an engineer, not a visionary. Nonetheless, I will claim that all music that comes after him is just a footnote to his work.

Anyone who has studied the Moonlight sonata, or the Fifth Symphony will recognise Beethoven's talent in engineering variations on a simple theme, which is a mathematical skill, but is there really a melody in the symphony?

As for Mozart...everyone knows how much melodic variety there is in his music.

How will one construct melodies?

Well here's an algorithm for constructing them, and infinite variety is already possible.

1. Write an eight-twelve-sixteen bar cantus firmus . There are already well know rules for writing one, but if one wants to write longer pieces of cf, it's always possible to add independent segments that are symmetric inside themselves, and connected at one end to the main line.

2. Now harmonise the cantus firmus, according to rules of counterpoint.

Counterpoint is a clearcut algorithm, but it may be rid of its contrapuntal character later, by appending harmonic fillers at points in accordance with the style that one wishes to compose in.

3. Now write a theme.

Alternatively....

2a. Write a theme on to the cantus firmus.

This will be the theme that will reappear on harmony notes of cf; it will keep appearing on the third and fifth of the phrase to provide coherence to the melody.

The theme can be random, as long as it remains bound by the laws of rhythm. Any theme can be made to make any emotional impact with appropriate dynamics, accidentals, tempo, and of course, performance. Reference Mozart...

4. While writing the melody,

a. Do not jump at the same pitch twice.

b. Change the register and direction between or inside each bar for climax.

c. Use one triad, based on the cantus firmus below, to create melodic elaborations based on the harmony. As long as melodic symmetry is preserved, a completely random rhythmic progression always succeeds. For example, if the cantus firmus is c, then our bar will travel between c and g, but the rhythmically accented points will always be on harmony notes and sevenths, and while moving to the next bar coherence will be achieved either by symmetry inside the beat, or by keeping the direction constant throughout successive bars. Of course this also has its own physhological effect, which may be manipulated to produce the desired effect.

5. Discard the cantus firmus. Elaborate on the elaboration to create a larger work.

It is possible to write complex, non-chordal melodies in fifteen minutes with this method by hand, but I admit that it's hard to make use of it as described here (I'm not sure that anyone will be interested) and it goes without saying that one has no idea on how they will sound untill he plays them, but it nonetheless succeeds. A computer can churn out at the pace of a genius, in case one takes the trouble to program it.

*skims over huge post*

*notices little/no useful or interesting information*

*leaves*

Wow. It's about time I found a good reason why people should stop trying to make music in the style of the original dudes. You're not going to be Mozart. You live in 2007. Do something that hasn't been done before.

Has anyone here heard of danger music?

This is a type of modern music I find most amusing and sometimes dumb.

Has anyone here heard of danger music?

No.

An older friend of mine whos in college was describing it to me last night.

(I dont know it very well but I am gonna get into it more)

It is music that inflicts pain upon the listener or the player.

For example: They take False Sehto(didnt spell it right) singer and have him scream into a mic. as high and as loud as he can. They then record it.

It is meant to be played only at a volume loud enough to inflict pain on the listener.

Modern enough? Some of it is interesting. Some of it is stupid though as well. Pls check it out.