Emmily Howell - scary future?

[Milan Simovski]

you guys might want to check this out

http://www.slate.com/id/2254232/

isn’t whether computers have a soul, but whether humans have a soul? will real composers be needed in the future?

read the whole post and please share your opinions. this kind of freaked me out, but its kind of interesting and creepy, imagine it in 10 years

listen to it too

[Milan Simovski]

very true!
I was saying about the composing not playing it... lets say you got a job for composing a music for some movie, and that same job is given to a software like this, it will do it in a few minutes all finished with score. And the other thing about it, all the emotions and inspiration, is it really all just "wanting to recognise music" rather than "hey I have an idea for a theme". Are we just biomachines wanting to recognise what we expirienced generations before? is it written in our code?
because this man proved that with a good programming and storing informations, repetition of the same stuff that made us cry. smile, happy, sad, repetes the process and we like it or not.
Do people tend to recognise rather than expirience?

about what you said I tottaly agree nothing can stop people for composing for the sake of composing.

Ray Kemp said:

For those of us who compose simply because we want to, this is totally irrelevant, we'll do it anyway.
For the music industry, it is also irrelevant as many young composers already give their music away for nothing.

Tell me how many samples you would need to play a virtuoso violin part that would get anywhere near matching the human player?
Too many.

Same for a saxophone. How many facial muscles, tongue movements etc etc.
Too many.

And on and on and on.

What I'm saying is.........I and probably most people want to listen to a performance including the odd flat or sharp note or dare I say even a wrong note, not a super piano roll player.

[Andrew Gleibman]

Don't be afraid Milan,

Just like a music box, even so sophisticated as Finale, does not express real human emotions (and we feel that when we listen to it however good are sample libraries), no computer program can come to something new musically - I can assure you as a professional AI researcher. Mass production of loud human-created musical scrap, bursting from moving cars, worries me more.

Maybe you heard of program "Eliza", made by Weizenbaum many years ago (see http://en.wikipedia.org/wiki/ELIZA ). Some ladies spent hours talking to this program: they claimed that it is the best "person" to communicate with they ever met. More sophisticated modern programs of this kind tell aphorisms and wise proverbs, extracting them from a "databases of human wisdom" simply by association with keywords contained in the user's phrases. More practical programs of this kind analyze your email texts and produce advertisement of related products near your printed text.

On the other hand, a properly structured database of music can be very useful. Students may study music using theme associations. Composers may immediately find music citations by printing short themes, chord sequences etc.

Now about composing automatically. About 35 years ago AI researchers did the following experiment.
They encoded 80-100 songs of a certain genre. Then a program produced new songs of this genre using
the following simple algorithm:

1) Take 4 bars of a song X (chosen randomly) and print them.
2) Find song Y, which contains similar 4 bars, and print next 4 bars from this song.
3) Assign Y as X and go to step 2 for forming next 4 bars (and so on, 10-20 times).

They found that they generate "new" songs of this genre. They carried out this experiment with many genres. But this was not composition. This was simply a research, aiming at better understanding what is a genre and how songs are structured, not more. Algorithms (even the most sophisticated) can only extract, combine and modify the stored traces of our emotions, they cannot produce new emotions.

[Milan Simovski]

yes indeed, and this eliza is very interesting too :)

[Nick Capocci]

Fear not! All the noise generated by all the computers in the world - whoever programmed them – will never amount to a single note of music. Fortunately for us, most people – trained or untrained – instinctively know this.

[Streaker Ofinsky]

In my view, writing the computer program is composing in a case like this. So human composers are still needed. You're simply setting up a framework where a specified range of indeterminacy will come into play to produce the final pieces. You can incorporate indeterminacy in similar ways when you're only dealing with humans, and that can result in many different-sounding pieces. That's nothing new, of course.

[Nick Olman]

Being a composer and a software developer, I find this idea quite intriguing. The one thing to always remember about programs is that they need everything to be broken down into the finest and most specific set of rules in order to adequately perform a function. For music this would include every theory around dynamics, timbre, melodic flow, chord structures and progressions, tempos, etc, in every possible permutation. The program will always follow logic similar to "if this is followed by this, then do these things". What it does can vary, as there are almost always options with our rules, so how would it know what the best one is? It can't exactly try it and see if it likes it. It doesn't know the difference between good and bad music from our perspective, it can only analyze patterns and make assumptions using a path of least resistance.

I do feel though, that this type of a program could be a great tool for composition. We've all run into less inspired times where we continuously stare at a page or computer screen with absolutely no clue how to start a piece. We could use a program such as this to create jumping points. Something to work from and around. If it really is true that all of our melodies come from a giant database of music we've already heard, wouldn't it be what we create around those ideas that makes us true composers?