More on Music AI
Music AI can't be all bad or boring, and may in fact be generative, which is one of the reasons music is interesting in the first place: there are smaller simple elements that can be combined in many ways. When you factor in the use of timbre, recordings and samples, data sets, and so on, it extends the possibilities even further.
This is an interesting use of a random generation of music, using the Lydian mode. As some musicians know the Lydian mode has an interesting "mathematical" facet, which arises from a series of stacked fifths. (See Lydian Chromatic Concept). Other symmetries exist in equal-temperament, such as the whole-tone scale and diminished scales in that they set up unresolved tensions seeking resolution. It wouldn't be difficult to set up an AI that would use these symmetries, then randomized by various other real-time data, such as weather or market data, but it might be boring, or one would have to sample it to find interesting segments that can work as listenable music.
This is an interesting use of a random generation of music, using the Lydian mode. As some musicians know the Lydian mode has an interesting "mathematical" facet, which arises from a series of stacked fifths. (See Lydian Chromatic Concept). Other symmetries exist in equal-temperament, such as the whole-tone scale and diminished scales in that they set up unresolved tensions seeking resolution. It wouldn't be difficult to set up an AI that would use these symmetries, then randomized by various other real-time data, such as weather or market data, but it might be boring, or one would have to sample it to find interesting segments that can work as listenable music.
