Stanford CS236 - Multi instrument MIDI music generation using Variational Auto Encoders
Recent advances in Generative models for Music gen- eration have shown impressive results. Learning the distribution of music creation may be an intractable problem at the moment. However one approach we can take is to build tools that serve as an aid in the creative process. If a musician already has a few ideas in mind on how a song or melody should start, can the problem be modelled as a conditional generative process where given the start can a model generate multiple possibilities of how the song can proceed?
SoundCloud Links:
https://soundcloud.com/saravana-r-389...