Don't worry, RNN generated version doesn't even generate a snippet of music. Calling it music is being generous, at best it is noodling. On the other hand, Mar

Don't worry, RNN generated version doesn't even generate a snippet of music. Calling it music is being generous, at best it is noodling. On the other hand, Markovian Models can segment and learn grammars. (Specifically probabilistic context free grammars.) The problem with these is overfitting.