No tags yet.
The Markov user interface device is working pretty well now, but there's one major problem with it - the random generations go on forever, representing the probability table and its representative Markov chain into infinite! It sounds chaotic, and although it is usually obvious that the melodies match the probability table, the endless stream of notes are not structured into catchy motifs. Music relies on motifs and patterns, and eletronic music even more so - these motifs and patterns thrive on repetition and gradual, subtle development.
My supervisor suggested that what I needed was a way of seeding the randomness - that way, I could refer back to a 16 note seed, and it would always be the same succession of notes. This seemed like such an obvious idea that I hadn't realised, and I instantly knew it was going to make a huge difference. However, the problem I am now having is that the [prob] object does not support the function of seeding in the same way that, for example, the [random] object does. I am going to have to find a way of either seeding the prob object, or making my own Markov object that does support seeding.