Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Vertigon]

Bah!  What reason is that not to go and do it over again!?  Besides, there's not enough games that actually *use* procedural music.  Actually, I can't think of any, of the top of my head.  Some of the CA stuff would be just about perfect as background music for certain games.

I don't think it's been mentioned yet, so Portal 2 uses procedurally generated music.

Derpedit: It has in fact been mentioned. Welp, I now have nothing to contribute to this thread :I

[Neonivek]

Yes but what does that mean Vertigon?

[Vertigon]

Which part, the not having anything to contribute or the procedurally generated bit? Because I'm working on the first part and Google has the second part taken care of.

[Draco18s]

So here's a composition I find quite pleasant.

[Bricks]

I'm always impressed how a simple change in key or tempo can drastically alter the interpretation of a piece.  I imagine the music for Cult is going to be more ambient and less incidental, since turn-based games don't really lend themselves to fixed paces.  Thus the whole Markov-chain thing really does seem like the way to go.  I made my own language generator (horribly inefficient but eh) using the same logic and was actually kind of disturbed how my brain tried to attach meaning to the pseudo-English words I generated (or the actual English words it came up with that weren't part of the source material).  I've even been thinking about ways to apply Markov chains to other procedurally-generated content, like items, monsters, and levels; possibly linking all those bits together so regions have distinct texture.

[EfTwelve]

I hope it's not too OT, But I wrote a program to do procedurally generated songs (http://www.kerneltrick.com).
Procedural music doesn't have to sound like jazz musicians stepping on geese. 

[Draco18s]

I hope it's not too OT, But I wrote a program to do procedurally generated songs (http://www.kerneltrick.com).
Procedural music doesn't have to sound like jazz musicians stepping on geese.

The output is OK, but you really need to work on smoothing the waveform of each note.  It looks sounds like the data for each note is being truncated at a non-0 position (i.e. where the waveform has no amplitude), so there's a bit of clipping going on as the pressure abruptly changes.

(I knew that sound editing class was good for something!)

[Lemunde]

Well, here's my attempt at procedural music generation. http://youtu.be/rZXjjoJit2A

[Bookend]

I haven't seen this mentioned yet, so here goes: http://www.earslap.com/projectslab/otomata.

Otomata generates music using a system of cellular automata that produce noise when they bounce off the walls. I've found it to be really easy to make pleasing patterns with. It provides good inspiration, too. It's kind of like a cross between that square-clicking thing posted earlier and Lemunde's music generator (which is pretty sweet, btw).

[Lord Dullard]

So, I've had a bit of a crazy idea about using Markov chains to create procedurally generated music.

I've started by using the Python midi package to record all midi events in a .mid file as parts in a single list. I just now got this to successfully work 100%, and have tested it by porting it back out as a new .mid file (which is, of course, identical to the original - but the important part is that I got the recording and list-interpreting algorithms to work correctly).

I'm now going to use these lists to, hopefully, create Markov chains which will be used to create 'new' songs. In reality they'll just be cannibalized patterns from existing midi files, but there are a few very cool advantages about doing this.

1. It will hopefully sound more 'human-composed' than typical algorithmically generated music, since it will focus on cannibalizing 'real' music instead of creating truly computer-generated sounds.

2. It will allow users to alter the types of music they want to generate by simply replacing/augmenting the pool of midi files.

3. It will allow a near-infinite amount of variety. Aside from being able to replace the midis themselves, I'm also going to try to get the program to mix and match Markov chains from *different* songs. This might get a bit wonky when dealing with songs in different keys/tempos, so I'm going to have to give some thought as to how I want to approach the problem.

I'll make any updates to progress. If I can get this to work, I plan on throwing it into Cult with a variety of options like saving favorite randomly generated tracks and assigning them to locations/characters, etc.

[freeformschooler]

Dullard, that is great news. I've always enjoyed procedural music, especially the theories behind it. Keep us updated.

[starsheep]

Not sure if it matters but after an year we made up a true procedural music game. It's called Music Invaders , you can know more at http://www.musicinvaders.com .
The concept behind it is that music can move object into the game , i'm not sure if you want me to explain the algorithmic part of it , but let's say that tempo , frequencies and beats are your friends if you want the concept to work.

[Mephisto]

I used to be big on trying out live coding with ChucK but I never got around to it. The ultimate in procedural generation - code the music yourself.