Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards This computer algorithm composes coherent symphonies

The subject is a direct quote, not my words.

At any rate, here's the link ...

I do have a small favor to ask. If you watch that video would you let me know if the video action stops or freezes while the sound continues. I've tried everything and the only way I can watch that entire video is to hover the mouse cursor over it and move the cursor from time to time.

... from that article ...

Professor Francisco Vico and a team of researchers at the University of Malaga in Spain created a computer program that can compose music. However, it doesn’t just create random noise that meets the technical definition of music, but can compose coherent symphonies that actually sound decent when played. In fact, it composed an entire album that was performed by the London Symphony Orchestra.

The program, called Iamus, has — perhaps oddly — composed over one billion songs, which will be put on sale free of royalties. The system used to create the music, as Vico calls it, is melomics. Though humans programmed the system, Iamus can run on its own afterwards, creating music without human input.

Interestingly and perhaps deviously, Vico noted in an interview with the Huffington Post that his software could be used for subliminal advertising purposes. He was able to reverse engineer the musical structure of the iconic Nokia ringtone, and his software made one million variants of it. Vico said this could be used as subliminal advertising — hearing something that isn’t actually a jingle, but still reminds you of that jingle.

Vico also envisions Iamus being able to create music that fits your mood. Rather than feeling energetic-but-secretly-sad today and then loading a ska playlist, you could tell Iamus your mood, and it could create original music to fit that mood. Coupled with some kind of mood-detecting app, Iamus could change the kind of music it creates to make sure you still aren’t listening to ska when you’re trying to fall asleep.

As for the future of Vico’s melomics, he feels that by the end of this year his software will be creating music that people won’t be able to differentiate from something made by a human. As cool as that sounds, we’re not looking forward to when bands start protesting about how robots took their jobs.

No great trick... Listen to the radio, it already sounds like a lot of current music is made by computer rather than human.

So it's a bit like a musical version of this poetry generator?:

http://thinkzone.wlonk.com/PoemGen/PoemGen.htm

Yes, it makes poems, but do they make any sense? Or would you want to read more than one?

That trio didn't move me . . . .

I did a music theory course in which the crowning glory was we had to compose a piece that was harmonically correct, followed rules about the base chords and their voicing, also rules about chord progressions, key changes, cadences, phrasing and form and so on. I did pretty well and got good marks. The piece sounded like cr** though!

David Cope has also made a very big thing of this:

David Cope

Whether it is difficult or not, a good imitation or not, are actually very complicated questions. Chopin would be a lot harder to do, I suspect, than a pretty good fugue. Ragtime would be much harder than some sorts of jazz. It is quite possible to write a small heuristic algorithm which produces reasonably interesting and varied fugues, for instance, I did that myself years ago. However, the professors in question do not seek heuristic means, but an overall intelligence coupled, in the case of imitation, which most of then are, with statistical analysis of pieces in known styles.

There was a thread here years ago, can't remember when, where readers tried to tell the difference between rough improvisations and serious modern compositions. No one could, but for some odd reason, perhaps more interesting than the result, the thread received severe criticism from some good musicians. So there is also the complication of the psychological aspect too, regardless of cleverness or quality.

For myself, the origin of a sound I like is irrelevant - might be spontaneous improvisation, carefully thought out composition, product of an algorithm or a chimpanzee - doesn't matter because all I care about is the sound. But as the article in the first post implies, most people do not think like this, and social, academic, historical and economic associations are very important. Hence the perceived threat.

A few weeks ago, in a science program series called 'Dara O'Briain's Science Club' on BBC TV (with James May among others, for the British people here), two fugues were played by a human on a harpsichord. One was composed by a computer following an algorithm, obeying all the rules of good composition and fugal writing, in the style of Bach. The other was genuine Bach. Both sounded very Bachian, but the musicians among the presenters could easily tell which was real Bach (as could I): the computer-generated fugue was just too straightforward, almost predictable, lacking any surprising element. A bit like a clever but studious student obeying all the rules.

The genuine Bach on the other hand went off on a little side twist, and there was also a strange modulation into a distant key, very briefly. It's the difference between a student and a master.

I guess it depends on what affect the computer was going for here, but it came out as someone who knows nothing about music. I don't feel threatened at all, woohoo!

Yup, this thread came to mind immediately upon reading the OP here.

Me and Sam were ones of the loudest voices against that thread...

My beef was simple: It was a trick thread to portray contemporary classical music in a very certain light and I didn't appreciate that at the time and I still don't. The idea was that ultimately all were impros and since people couldn't tell it was obvious that all contemporary classical music is rubbish and indistinguishable from random impros. No matter the study or the effort, etc. Lame!

Now, onto this.

I've seen it, I've checked it and I understand the philosophies behind it. In fact my PhD thesis had a lot of comments about Dave Cope (Iamus was not out back in 2009-2010, or at least I didn't know about it).

If you think about it, there's already a huge amount of 'premade' music into our ipods, heads, etc. All band in a box, garageband, loops, samples, etc do exactly that: Offer a premade phrase, or part, or idea to us. Even the simple casio keyboards, offer automatic accompaniament and this includes drums, bass, and some other instruments.

So there's every reason to believe that ultimately Iamus and other software will work better, or good enough for the general audience to take an interest. And, as it's happening with sample libraries and computers, etc, nobody lost their job from sample libraries: The orchestras are still getting used. Samples are used simply when the budget is not enough!

David Cope's work on computer-assisted composition is sophisticated, extraordinary - and controversial - on so many levels. That's not to say his programs churn out music we all look to hear to in a great concert - because they don't. His work, however, does raise genuine and difficult questions about what it actually means to "compose" and to be "inspired" and even "to be human." Those, in particular, are areas of knowledge that frustrate many of his critics who would often judr rather assume that the answers to these kinds of questions are obvious. Cope usually refers to his work as "computer-assisted" rather than "computer-generated."

exactly right, computers have been composing already music for a while now.

Computers are used to produce todays music, not compose it. Important difference. Even if all that simplistic stuff you hear in the charts sounds like someone just pressed some buttons, the composing was done by humans.

(edit) I forgot to say that the video works fine.

Do any of you guys like what is played in the video?

Oh dear, I misconstrued the point of that old thread altogether then. I interpreted it as implying that all that matters to a listening mind is the sound, regardless of how it originated. The idea that the poster was specifically attacking contemporary composition as a particular means of creation of that sound never occurred to me. Perhaps I didn't read carefully enough. On the other hand, my naivety could be a blessing, as subtle negativity frequently goes in one ear and out the other. Thanks for pointing it out.

Have you tried out Band-In-the-Box lately Dave? It can churn out pretty good cliché jazz solos. I was actually amazed at how well it handles standard progressions. The solos are well laid out with a coherent shape.

But it doesn't sound right to me when it does more complex tunes. I guess it's trying to force Bebop licks in which kind of explains the logic of the programmer.

Here's an example. To my ears, this is not in keeping with the tune.

Inner Urge Solo by Band-In-The-Box
https://www.box.com/s/ers62eycuqf70lz05cdq

Where it fails is that it doesn't know how to go outside of the harmony to use common tones to extend melodies. Instead it looks for a progression and then I suppose looks for a collection of licks that fit the progression. Since this tune has a non-functional progression, I guess the logic tries it chord by chord. So it sounds disjointed.

But in any case, it's interesting how it's done. And it also means that as humans, we can play in a formulaic way with no artistry if we so choose.