Monday, 16 May 2011

Crowd Sourced Blues Generation.

"I woke up this morning.. my blues were all red, 
 Been thinking about, generation instead..
 Chatting with Gandalf, putting thoughts in my head,
 That blues is a sequence, of riffs in a thread.
 Woaah-oh.. maybe you could get some, of my gener-ated-blues.. "

Ok, so, when noodling blues, I have a whole bunch of little tiny riffs, that I glue together based on where I think the music should go next, and when you stop to think about it a bit, the sequencing isn't entirely random.

Perhaps there's enough structure present within the music to allow construction of a system where anyone can contribute to the music, while the system acts as composer & arranger..

Firstly, you know that some riffs naturally lead into others, and others that just wont..

Secondly, the history of what's been used already may cause the re-prioritisation of some riffs, to create patterns through repetition, or similar sounds. (eg. Chorus, Verse, Bridge, Solo)

This kind of leads to a concept of a directed graph, where each Node represents a riff, with connections to others, which are weighted based on current location, and on the breadcrumb trail of nodes past visited.

Imagine then if the nodes could be crowd sourced, just for example, by adding them as comments to a discussion thread.. and that the participants of the thread can up-vote/down-vote on new nodes (This to help weed out the inevitable 'chopsticks', or 'cat walking on the piano' trolling nodes). Think of something like Slashdot where comments are voted up / down, with users having karma to distribute by way of votes, but instead of comments, having short riffs.

At first thought, the node exit/entry criteria would be part of the node.. so the node author would have some control over how the flow of the music would go. But with deeper thought, perhaps the criteria should be separated from the node, made part of a set-overlay, enabling construction of 'themes' where the same riffs are utilised, but with different connecting weightings producing a different feel to the output.

If such an imaginary system were to exist, and were to span a wide enough timeframe, it may also be interesting to track how a given collection of nodes evolves over time, would it's changes represent the changing sound of music over time? or would it merely become more expansive, refined than its seed set.

No comments:

Post a Comment