It is currently Thu Mar 28, 2024 1:35 am




Post new topic Reply to topic  [ 12 posts ]  Go to page Previous  1, 2
 Creativity 
Author Message
Master Contributor
Master Contributor
User avatar

Joined: Wed Sep 19, 2012 8:35 am
Posts: 1210
Location: Indonesia
Post Re: Creativity
There's this music too:
https://www.youtube.com/watch?v=RIawYiX3D84

Notice how it's titled Dingularity. That seems to be a slap in the face to Kurzweil's singularity. Thaler talked about jealousy in one of his other articles so I think there's some rivalry. Maybe he means Kurzweil is a Ding Dong. Haha

He's also got videos on youtube explaining his creation.

_________________
www.soundclick.com/gongchime


Sun Sep 28, 2014 2:10 am
Profile
Master Contributor
Master Contributor
User avatar

Joined: Wed Sep 19, 2012 8:35 am
Posts: 1210
Location: Indonesia
Post Re: Weighted Composition Scheme
Interval and pitch are only 2 of the parameters you could “degrade” toward innovation using the ideas from the article on artificial neural nets. The computer isn’t required to compose like in the second article where it isn’t given any data except constraints, and it pulls itself up by its bootstraps as it were by analyzing your facial responses to what it comes up with until it finally comes up with something passable like “Dingularity.” ---

If I understand correctly that “Crying in a Vacuum” was composed by first inputting the creator’s favorite pieces and then degrading some parts from each one and recombining only a few parts from different songs, we can follow in his footsteps as an experiment. ---

Actually, human songwriters and composers do the same thing. We learn a bunch of songs and then think, “I can do that,” and start to copy our favorite bands to find out just how hard it is. Brahms was of the opinion you have to compose 100 songs before you’ll write any good ones. ---

Also, the normal system that composes using fractals and genetic markov chains are hit and miss. Even when they finally put out something which is mostly good, there’s usually still at least one fatal flaw that makes it sound like something’s wrong. Only when combined with genetic algorithms does that problem start to go away. ---

The other method that I talked about before where an improviser’s line is transformed to create music in real time is remarkably similar to the method of Thaler’s neural net. In that method, the original melody is “degraded” using musically meaningful variation procedures like transposition, augmentation, diminution, retrograde, inversion etc…Based on Thaler’s work, the question becomes how many times would you have to do this before it was altered 6/10ths from the original? If the melody had 10 notes, you could alter 6 of them. But then how far to alter them is another question. 6/10ths of the 12 chromatic pitches is 7.2 so you might think to move every note about a perfect fifth. But you don’t want every melody with 10 notes to have exactly 6 notes altered. And you don’t want to move every note that should be moved the same interval of a 5th. ---

I’ve never been looking for more of the status quo in pop, rock or jazz. I’m looking for highly innovative new music. So, normally you’re supposed to make the lengths of all the parts the same. This is probably one of the constraints programmed in Thaler’s bootstrapping approach. Sections should be 16 ms too but if you’re allowed to degrade even basic parameters which makes everything congruent only 6/10ths of the time, that would be very novel. I have some experience with these kinds of schemes so I’m giving you easy examples here on how to do some of this stuff. ---

A long time ago I found some data on pitches and intervals and I deducted the data for rhythms so have been working with weighted pitch, interval and rhythm sets for a while. The data I found says how often a pitch or an interval gets used. For the C major scale, out of thousands of songs, C and G have an equal weight of 8. E and B have an equal weight of 6. D and A have an equal weight of 4. And F and high C have an equal weight of 2. ---

If you add those all together you get a total of 20. We want to change these pitch weights to look for innovation. 6/10ths of 20 is 12. So, if we count past the 8 assigned to C and G which currently have the highest weight and into E and B we stop on 12 which is over halfway through the 6 of E and B. What we want to do now is make E and B get the highest weight and reduce the weight for C and G. To do that we’re just going to shift those numbers so that E and B get the higher number which is 8. This makes D and A change to 6, and F and high C get 4. C and G are still very important though not as important as before so we give them a 6 just like D and A. ---

Now they add up to 24 total so go get 24 cards and write C and G on 6 of them, E and B on 8 of them etc… Then you can draw cards stochastically (not randomly since they’re weighted) to decide a pitch set or series. --

For intervals I’m leaving out unisons and octaves. They’re normally used in special places like between phrases or during the climax. I don’t want them randomly showing up all the time in the middle of every other phrase. I see where other computer music programmers do a process they call concatenation where multiple instances of a note back to back are reduced to a single instance, so I feel justified in taking this approach. --

The numbers for intervals were too big so I’ve reduced them equally for each interval. The lowest was 13 for the tritone so I divided all the numbers by 13 so now the tritone gets a 1. So from largest to smallest the weights for the interval of a 2nd = 234, 3rd =79, 6th=49, 7th=28, 4th=26, 5th=19, and the tritone =1. Adding all these together gives us 436. 6/10ths of 436 is 261.6. We count past the 234 of the 2nds and not quite half way into the 79of the 3rds. So we can see the 3rds should get more weight if we’re looking for innovation. Again, just shift the numbers over. Now the thirds are going to be given the greater weight with 234, 6ths=79, 7ths =49, 4ths=28, 5ths=26, TT=19. The 2nds will be a bit less important than before so now are given 79 just like the 3rds. ---

I can give rhythms too if you’re interested.

_________________
www.soundclick.com/gongchime


Sun Sep 28, 2014 7:55 am
Profile
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 12 posts ]  Go to page Previous  1, 2


Who is online

Users browsing this forum: No registered users and 12 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron

board3 Portal - based on phpBB3 Portal Powered by phpBB © phpBB Group. Designed by Vjacheslav Trushkin for Free Forums/DivisionCore.
Heavily modified by Stickist.com. Stickist.com is an authorized Chapman Stick® site. The Chapman Stick® and NS/Stick™ and their marks are federally registered trademarks exclusively licensed to Stick Enterprises, Inc., and are used on Stickist.com and NSstickist.com with SEI's permission.
Click here for more information.