Citation
Sophie Libkind, Alexandre Donze, Sanjit Seshia, David Wessel. "Control Improvisation with Applications to Music". Talk or presentation, 29, September, 2013; Presented at First International Workshop on the Swarm at the Edge of the Cloud (SEC'13 @ ESWeek), Montreal.

Abstract
Control Improvisation is the process of generating control events in a creative way while accommodating a set of given specifications in a changing environment. We study this problem in the context of musical improvisation which aims at producing musical material on an automatic and incremental fashion. More specifically, we consider the scenario of a monophonic melody (solo) generation on a given jazz song harmonization. The generated music is symbolic, in the sense that our improviser creates sequences of notes symbols, each specifying a pair of a pitch and a discrete duration. Our approach can be decomposed roughly into two components: one generator that learns from a training sequence (obtained from a human improviser or externally generated) and outputs "similar'', although randomized in some way, melodies, and one supervisor which sets constraints on the generator so that the melodies respect specifications both in the pitch and in the rhythmic domain. We use a measure inspired from Normalized Compression Distances (NCD) to estimate the divergence between generated melodies and the training melody and propose an algorithm to formally control this divergence.

Electronic downloads

• ControlImprovWithAppsToMusic_Libkind_SEC13_Montreal.pdf · application/pdf · 370 kbytes
• http://www.src.org/library/publication/p067997/. (SEC Website)
• libkind_improvmusic_edge.pdf · application/pdf · 147 kbytes. (Abstract)

• HTML

  Sophie Libkind, Alexandre Donze, Sanjit Seshia, David
  Wessel. <a
  href="http://www.terraswarm.org/pubs/110.html"><i>Control
  Improvisation with Applications to
  Music</i></a>, Talk or presentation,  29,
  September, 2013; Presented at <a
  href="http://www.terraswarm.org/conferences/13/swarm/index.htm"
  >First International Workshop on the Swarm at the Edge of
  the Cloud (SEC'13 @ ESWeek)</a>, Montreal.

• Plain text

  Sophie Libkind, Alexandre Donze, Sanjit Seshia, David
  Wessel. "Control Improvisation with Applications to
  Music". Talk or presentation,  29, September, 2013;
  Presented at <a
  href="http://www.terraswarm.org/conferences/13/swarm/index.htm"
  >First International Workshop on the Swarm at the Edge of
  the Cloud (SEC'13 @ ESWeek)</a>, Montreal.

• BibTeX

  @presentation{LibkindDonzeSeshiaWessel13_ControlImprovisationWithApplicationsToMusic,
      author = {Sophie Libkind and Alexandre Donze and Sanjit
                Seshia and David Wessel},
      title = {Control Improvisation with Applications to Music},
      day = {29},
      month = {September},
      year = {2013},
      note = {Presented at <a
                href="http://www.terraswarm.org/conferences/13/swarm/index.htm"
                >First International Workshop on the Swarm at the
                Edge of the Cloud (SEC'13 @ ESWeek)</a>, Montreal.},
      abstract = {Control Improvisation is the process of generating
                control events in a creative way while
                accommodating a set of given specifications in a
                changing environment. We study this problem in the
                context of musical improvisation which aims at
                producing musical material on an automatic and
                incremental fashion. More specifically, we
                consider the scenario of a monophonic melody
                (solo) generation on a given jazz song
                harmonization. The generated music is symbolic, in
                the sense that our improviser creates sequences of
                notes symbols, each specifying a pair of a pitch
                and a discrete duration. Our approach can be
                decomposed roughly into two components: one
                generator that learns from a training sequence
                (obtained from a human improviser or externally
                generated) and outputs "similar'', although
                randomized in some way, melodies, and one
                supervisor which sets constraints on the generator
                so that the melodies respect specifications both
                in the pitch and in the rhythmic domain. We use a
                measure inspired from Normalized Compression
                Distances (NCD) to estimate the divergence between
                generated melodies and the training melody and
                propose an algorithm to formally control this
                divergence.},
      URL = {http://terraswarm.org/pubs/110.html}
  }
  

Posted by Christopher Brooks on 29 Sep 2013.