eduzhai > Applied Sciences > Engineering >

Generating Music with a Self-Correcting Non-Chronological Autoregressive Model

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 8 pages

Abstract: We describe a novel approach for generating music using a self-correcting,non-chronological, autoregressive model. We represent music as a sequence ofedit events, each of which denotes either the addition or removal of anote---even a note previously generated by the model. During inference, wegenerate one edit event at a time using direct ancestral sampling. Our approachallows the model to fix previous mistakes such as incorrectly sampled notes andprevent accumulation of errors which autoregressive models are prone to have.Another benefit is a finer, note-by-note control during human and AIcollaborative composition. We show through quantitative metrics and humansurvey evaluation that our approach generates better results than orderlessNADE and Gibbs sampling approaches.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×