Ghent University Data science seminar: Dr. Sander Dieleman

(22-02-2019) We were excited to welcome back Dr. Sander Dieleman (Google DeepMind) at ELIS for the #DataScienceSeminar. He spoke about the challenge to generate realistic sounding speech or music.

Realistic music generation is a challenging task. When machine learning is used to build generative models of music, typically high-level representations such as scores, piano rolls or MIDI sequences are used that abstract away the idiosyncrasies of a particular performance. But these nuances are very important for our perception of musicality and realism, so we embark on modelling music in the raw audio domain. I will discuss some of the advantages and disadvantages of this approach, and the challenges it entails.

 

More information