MENU

Dr. Gaëtan Hadjeres

Associate Researcher

music

Putting composers back in the loop

Applying the latest deep learning techniques to music composition is appealing for AI researchers; but for composers, this intrusion of machines in their domain of expertise could be perceived as a threat. This fear of being replaced is legitimate: indeed, many recent generative models for music tend to produce infinite numbers of
scores without the need for human intervention. I think that this behavior is not desirable and that AI algorithms should instead be used by artists as assistants during the compositional process. By creating a fruitful discussion between a composer and the machine, the artist can then focus on the development of their musical ideas and let the AI do the technical parts. Professional composers can benefit from these tools to become more productive and explore uncharted regions of musical creation while amateur musicians can use these innovative tools to express themselves in an intuitive way. By putting composers back in the loop, we will go from automatic music composition to AI-augmented composition and redefine the way people compose music.

2019

Monte Carlo Information-Geometric Structures

Topics:
music
Authors
N. Frank , G. Hadjeres |

Geometric Structures of Information, edited by:Nielsen, Frank, Springer International Publishing, Cham, 2019. pp.69--103.

read more

Abstract

Exponential families and mixture families are parametric probability models that can be geometrically studied as smooth statistical manifolds with respect to any statistical divergence like the Kullback–Leibler (KL) divergence or the Hellinger divergence. When equipping a statistical manifold with the KL divergence, the induced manifold structure is dually flat, and the KL divergence between distributions amounts to an equivalent Bregman divergence on their corresponding parameters. In practice, the corresponding Bregman generators of mixture/exponential families require to perform definite integral calculus that can either be too time-consuming (for exponentially large discrete support case) or even do not admit closed-form formula (for continuous support case). In these cases, the dually flat construction remains theoretical and cannot be used by information-geometric algorithms. To bypass this problem, we consider performing stochastic Monte Carlo (MC) estimation of those integral-based mixture/exponential family Bregman generators. We show that, under natural assumptions, these MC generators are almost surely Bregman generators. We define a series of dually flat information geometries, termed Monte Carlo Information Geometries, that increasingly-finely approximate the untractable geometry. The advantage of this MCIG is that it allows a practical use of the Bregman algorithmic toolbox on a wide range of probability distribution families. We demonstrate our approach with a clustering task on a mixture family manifold. We then show how to generate MCIG for arbitrary separable statistical divergence between distributions belonging to a same parametric family of distributions.

2019

Variation Network: Learning High-level Attributes for Controlled Input Manipulation

Topics:
music
Authors
Gaëtan Hadjeres |

CoRR, abs/1901.03634, 2019.

2019

Neural Drum Machine : An Interactive System for Real-time Synthesis of Drum Sounds

Topics:
music
Authors
Cyran Aouameur , Esling Philippe , Gaëtan Hadjeres |

arXiv preprint arXiv:1907.02637, 2019.

2019

Learning to Traverse Latent Spaces for Musical Score Inpainting

Topics:
music
Authors
Pati Ashis , Lerch Alexander , Gaëtan Hadjeres ,

arXiv preprint arXiv:1907.01164, 2019.

2019

On power chi expansions of f-divergences

Topics:
music
Authors
Frank Nielsen , Gaëtan Hadjeres |

CoRR, abs/1903.05818, March, 2019.

2018

Monte Carlo Information Geometry: The dually flat case

Topics:
music
Authors
Nielsen Frank , Gaëtan Hadjeres |

2018.

2018

Machine Learning Research that Matters for Music Creation: A Case Study

Topics:
music
Authors
Sturm Bob , Ben-Tal Oded , Monaghan Una , Collins Nick , Herremans Dorien , Chew Elaine , Gaëtan Hadjeres |

Journal of New Music Research, 2018.

2018

Anticipation-RNN: Enforcing Unary Constraints in Sequence Generation, with Application to Interactive Music Generation

Topics:
music
Authors
Gaëtan Hadjeres |

Neural Computing and Applications, special issue on Deep Learning and Music, 2018.

2017

DeepBach: a Steerable Model for Bach Chorales Generation

Topics:
music
Authors
Gaëtan Hadjeres |

Proceedings of the 34th International Conference on Machine Learning, edited by:Doina Precup and Yee Whye Teh, 70, PMLR, International Convention Centre, Sydney, Australia, August, 2017. pp.1362--1371.

2017

Sampling Variations of Sequences for Structured Music Generation

Topics:
music
Authors
François Pachet , Gaëtan Hadjeres |

Proceedings of the 18th International Society for Music Information Retrieval Conference, ISMIR, Suzhou, China, October, 2017. pp.23-27.

2017

Sampling Variations of Lead Sheets

Topics:
music
Authors
Pierre Roy , Gaëtan Hadjeres |

March, 2017.

2017

GLSR-VAE: Geodesic latent space regularization for variational autoencoder architectures

Topics:
music
Authors
Gaëtan Hadjeres |

2017 IEEE Symposium Series on Computational Intelligence (SSCI 2017), IEEE, 2017. pp.1--7.

2017

Deep learning techniques for music generation-a survey

Topics:
music
Authors
Briot Jean-Pierre , Gaëtan Hadjeres |

2017.

2017

Deep rank-based transposition-invariant distances on musical sequences

Topics:
music
Authors
Gaëtan Hadjeres , Nielsen Frank |

CoRR, abs/1709.00740, 2017.

2016

Style Imitation and Chord Invention in Polyphonic Music with Exponential Families

Topics:
music
Authors
Gaëtan Hadjeres |

September, 2016.

2016

DeepBach: a Steerable Model for Bach chorales generation

Topics:
music
Authors
Gaëtan Hadjeres |

December, 2016.

2016

Generating non-plagiaristic Markov sequences with max order Sampling

Topics:
music
Authors
Gaëtan Hadjeres |

Creativity and Universality in Language, edited by:Degli Esposti, M. and Altmann, E. and Pachet, François, Springer, 2016.

2016

Enforcing Structure on Temporal Sequences: the Allen Constraint

Topics:
music
Authors
Pierre Roy , Perez Guillaume , Régin Jean-Charles , Gaëtan Hadjeres |

Proceedings of the 22nd International Conference on Principles and Practice of Constraint Programming - CP, Springer, Toulouse, France, September, 2016.

2016

Assisted Lead Sheet Composition using FlowComposer

Topics:
music
Authors
Gaëtan Hadjeres |

Proceedings of the 22nd International Conference on Principles and Practice of Constraint Programming - CP, Toulouse, France, September, 2016.

2015

Generating 1/f Noise Sequences as Constraint Satisfaction: the Voss Constraint

Topics:
music
Authors
François Pachet , Pierre Roy , Gaëtan Hadjeres |

24th International Joint Conference on Artificial Intelligence (IJCAI 2015), Buenos Aires (Argentina), July, 2015.

2015

Approximating Covering and Minimum Enclosing Balls in Hyperbolic Geometry

Topics:
music
Authors
Nielsen Frank , Gaëtan Hadjeres |

Geometric Science of Information, edited by:Nielsen, Frank and Barbaresco, Fr{'e}d{'e}ric, Springer International Publishing, Cham, 2015. pp.586--594.

2014

Avoiding Plagiarism in Markov Sequence Generation

Topics:
music
Authors
Gaëtan Hadjeres |

28th Conference on Artificial Intelligence (AAAI 2014), Quebec (Canada), July, 2014. pp. 2731--2737.