Hi there—

I've started research for an interesting project that requires real-time generation of simple music for infants and children. I'm reading up on different implementations of Schenkarian-style music generation — or rather Schenkarian analysis done in reverse to create original musical pieces. It seems that Pure Data would be an easy way to accomplish this goal — would anyone be able to give me some advice on where to begin, or point me to other Pure Data projects that deal with the same topic? I'd like to explore as many different examples as possible before making my own attempt.

Thanks!

-Chris

One simple starting point could be to use randomly chosen notes from a list or a table. For example, you could have a "tonic" table containing of "0 0 0 2 4 4 4 5 7 7 7 9 10 11 12" and read it with a [random 15]. This will give you a weighted random that generates MIDI notes in a major scale, but focusing on the I chord. With another table for pre-dominant and dominant, you could get some function algorithmically.

I like the weighted random idea —

I think I'm going to start by implementing the bare-bones core structure of a piece, starting with the root note and mode, and then generating a chord structure based off of this mode. The idea of Schenkarian analysis is to start off by analyzing the larger chord structure of a piece (to find the overarching I-V-I motion) and then move inwards layer by layer in terms of complexity until the whole piece is analyzed down to the individual notes. I'm going to approach generation in the same way — when I get to the point where I am generating the smaller chord progressions I will probably use a weighted random scheme of commonly used motives.