Generative Melody Composition with Human-in-the-Loop Bayesian Optimization

Deep generative models allow even novice composers to generate various melodies by sampling latent vectors. However, finding the desired melody is challenging since the latent space is unintuitive and high-dimensional. In this work, we present an interactive system that supports generative melody composition with human-in-the-loop Bayesian optimization (BO). This system takes a mixed-initiative approach; the system generates candidate melodies to evaluate, and the user evaluates them and provides preferential feedback (i.e., picking the best melody among the candidates) to the system. This process is iteratively performed based on BO techniques until the user finds the desired melody. We conducted a pilot study using our prototype system, suggesting the potential of this approach.

[1]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[2]  Colin Raffel,et al.  A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music , 2018, ICML.

[3]  Takeo Igarashi,et al.  Interactive Subspace Exploration on Generative Image Modelling , 2019, ArXiv.

[4]  Marco Cuturi,et al.  Soft-DTW: a Differentiable Loss Function for Time-Series , 2017, ICML.

[5]  Maya R. Gupta,et al.  How to Analyze Paired Comparison Data , 2011 .

[6]  Douglas Eck,et al.  Towards Mixed-initiative generation of multi-channel sequential structure , 2018, ICLR.

[7]  Adrien Bardet,et al.  Flow Synthesizer: Universal Audio Synthesizer Control with Normalizing Flows , 2019, Applied Sciences.

[8]  Ching-Hua Chuan,et al.  A Functional Taxonomy of Music Generation Systems , 2017, ACM Comput. Surv..

[9]  Nando de Freitas,et al.  A Bayesian interactive optimization approach to procedural animation design , 2010, SCA '10.

[10]  Takuya Akiba,et al.  Optuna: A Next-generation Hyperparameter Optimization Framework , 2019, KDD.

[11]  Bernhard Schölkopf,et al.  From Variational to Deterministic Autoencoders , 2019, ICLR.

[12]  Antonio Torralba,et al.  Generating Videos with Scene Dynamics , 2016, NIPS.

[13]  Cheng-Zhi Anna Huang,et al.  Novice-AI Music Co-Creation via AI-Steering Tools for Deep Generative Models , 2020, CHI.

[14]  Jose D. Fernández,et al.  AI Methods in Algorithmic Composition: A Comprehensive Survey , 2013, J. Artif. Intell. Res..

[15]  Nando de Freitas,et al.  Active Preference Learning with Discrete Choice Data , 2007, NIPS.

[16]  Nando de Freitas,et al.  Taking the Human Out of the Loop: A Review of Bayesian Optimization , 2016, Proceedings of the IEEE.

[17]  Jesse Engel,et al.  Magenta Studio: Augmenting Creativity with Deep Learning in Ableton Live , 2019 .

[18]  Peter Knees,et al.  An Intelligent Musical Rhythm Variation Interface , 2016, IUI Companion.

[19]  Takeo Igarashi,et al.  Sequential line search for efficient visual design optimization by crowds , 2017, ACM Trans. Graph..

[20]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[21]  Yi-Hsuan Yang,et al.  MidiNet: A Convolutional Generative Adversarial Network for Symbolic-Domain Music Generation , 2017, ISMIR.

[22]  Curtis Hawthorne,et al.  Magenta.js: A JavaScript API for Augmenting Creativity with Deep Learning , 2018 .

[23]  Alexei A. Efros,et al.  Generative Visual Manipulation on the Natural Image Manifold , 2016, ECCV.

[24]  Penousal Machado,et al.  Deep Learning for Expressive Music Generation , 2019, ARTECH.

[25]  Monica Dinculescu,et al.  MidiMe: Personalizing a MusicVAE model with user data , 2019 .

[26]  Nando de Freitas,et al.  Bayesian Optimization in a Billion Dimensions via Random Embeddings , 2013, J. Artif. Intell. Res..

[27]  Issei Sato,et al.  Sequential gallery for interactive visual design optimization , 2020, ACM Trans. Graph..

[28]  Takeo Igarashi,et al.  Human-in-the-loop differential subspace search in high-dimensional latent space , 2020, ACM Trans. Graph..