Abundant Music: An online procedural music generator

Engines, Libraries and Tools Tools Discussion
#1

Hello there :slight_smile:
I have been working with generative music for a long time now and want to demonstrate my new fancy system to my fellow Java game constructors.
The idea is to use this “tool” to generate music for whatever purpose you like and be able to modify the structure, intensity etc. on a high level without having to modify everything from scratch.

Try it here!

Get started by reading some tutorials.

If you want to render or edit midi files yourself, I can recommend the free editor Synthfont.

Some youtube clips that shows some output from the system:

pV9RZEbfHGA

QHslgrhY8d8

I hope that this system will be used someday for generating music from within a game and not just offline. Currently, it uses planning and constraint solvers that takes some time. Each song is approximately composed in 2-5 seconds but can probably be 5-10 times faster.

Note that this music is not generated by the RML Editor, which is one of my previous attempts with generative/reactive music. This is a new, not published software that generates this.

#2

What do you mean generate music? These were made with a push of a button?? Are you serious??? Holy shit??!

#3

Yes, that’s correct :slight_smile:
It’s still very WIP though.

#4

Wow, this is incredible! By any chance will you be posting a download link for the executable? This would certainly spice up my games, and I would think pretty much any game. Just from clicking a few of the links, this is nearly as good as better than some music that was custom made.

EDIT: Found the link commented out on your websites main page. http://www.springworldgames.com/musicgenerator/
TL;DR Very impressive

#5

Wow. Two small requests:

  • Make some epic length tracks (like 10, 20 minutes) so I can get totally baked relax to them.

  • Put some of these on youtube and put a [ youtube ] tag in the post so people can “mash butan” and listen instantly.

#6

This is epic! Reminds me of Transport Tycoon and Theme Hospital, hehe. Even Final Fantasy VII, even though a generator can never beat Nobuo Uematsu, the master of the MIDI.

Too bad they’re MP3s. I’d like to try them out with my Yamaha XG SoftSynth S-YXG50 MIDI synthesizer on.

Amazing work you’ve done.

#7

Actually, that is an old version that doesn’t compose music the same way.
The current version uses constraints and planning based on voice leading “rules” while the old one uses no such things.

#8

Ahh, my apologies then. I saw the name and assumed it was it.

#9

Great ideas, thanks!

#10

I have the MIDIs available so you can get them if you like. I will put them at the same place but they will be called test1.mid instead of test1.mp3 and so on :slight_smile:

I wanted to show the mp3s first since they are rendered with a reasonably good soundbank and the Gervill synthesizer (which has some reverb and other nice things available).

#11

I’d like to be tester :slight_smile: (This fits into a terraria-minecraft2d-like generated sandbox game, from the point of view about the generated stuff)

#12

This is so epic! I made a little evening-project of it :slight_smile:
http://soundcloud.com/ultroman-the-tacoman/auto-generated1

Just played with one of your applets, and ran it through my mastering suite.

What I would love, is to see you put in some crashes. Particularly when there’s a significant change in intensity or harshness. Also, it seems intensity has an influence on many factors, including volume or velocity. From a producer’s point of view, I’d love to see volume and over-all velocity separated from intensity, if that is the case. If it only has influence on velocity, then nevermind. It just seems that many things change when I turn it up, and because a song usually doesn’t change that much in volume, I seem to be restricted from certain levels of intensity, because it has so much influence on the volume that abrupt changes make the transitions sound unnatural. I know it’ll probably mean a lot more automation-work for your generator, but I think it’ll make it more dynamic.

But brilliant work! This is…unbelievable.

#13

You’ve got a follower on tumblr :slight_smile:

#14

Great! That sounds cool.
It is actually from my old work using the RML editor. There you can actually see exactly what happens if you analyze the RML module, that you modify the parameters for.
You can also change the module with the editor so it fits your needs, if you can decipher the badly designed GUI :slight_smile:

The mp3s above are from the new, hopefully more sophisticated version where I try to use some proper voice leading with constraints and all that.

But the new version can not deal with real-time controllers for the music (yet?), so then the RML editor is the only choice at the moment.
It is also a good thing that the RML editor is available with source and all so you can just embed it if you want that and you find the result good enough.

#15

Brilliant work. With some runtime inputs for mood, this would be awesome :slight_smile: (does that sound greedy?)

#16

Wow, that’s my first right? :slight_smile:

#17

The system will be parameterized better later. It is currently just a box where you enter a single seed and get a MIDI out.

#18

SUBARASHEE! ;D another round-up tool for LD.

#19

Impressive !

#20

I think so :slight_smile:
You’re my second (lol…)