cage.markov: "unsticking" it

[Lauprellim]

Hi,

I've been building upon the basic help patch for [cage.markov.analysis] and I'm surprised at how often it gets "stuck", particularly when order > 1. I know there's a message you can send to the third inlet of [cage.markov.synthesis] ("clear") to "unstick" it, but I'm finding that I have to do this far too often. The more sophisticated help patch (which plays the Bach cello suite) doesn't get stuck as frequently.

Can you tell me a little more about why [cage.markov.x] gets stuck, what I can do to minimize the chances of it getting stuck, or maybe how to prevent it altogether?

A million thanks as always!

[Lauprellim]

Hi again:

Quick followup: I think part of the problem is that when I have a higher-order chain set, I might not provide the algorithm a "way back" to any of the stored states. Am I correct in thinking that the source of my problem is this: if [cage.markov.synthesis] gets to state x and there is no state y to go to next, it will get stuck?

[danieleghisi]

Precisely!
That's what every standard Markov chain does when it reaches a "final" state. Is this the case? Does the chain always end with a "final" token?

[Lauprellim]

Hi,

In the MIDI files I downloaded from kunstderfuge.com, there was often a final state when I ran them through cage.markov.analysis. Inputting pitches manually and providing a "loop" back makes the whole thing behave much better! These cage markov tools are remarkable and very useful. THANK YOU.