Keynote at DMRN on controllable music generation

I was excited to give a keynote at DMRN+15. More info on the workshop here. Listen to my keynote below!

Controllable music generation: from MorpheuS to deep networks

In its more than 60 year history, music generation systems have never been more popular than today. In this talk, I will discuss a number of co-creative music generation systems that have been developed over the last few years. These include MorpheuS, a tonal tension-steered music generation system guided by tonal tension and long-term structure. MusicFaderNets, a variational auto encoder model that allows for controllable arousal and rhythmic density of music. Finally, some more recent models by our AMAAI lab which include architectures such as controllable transformers and hierarchical RNN.