MINGUS: Melodic Improvisation Neural Generator Using Seq2Seq

Sequence to Sequence (Seq2Seq) approaches have shown good performances in automatic music generation. We introduce MINGUS, a Transformer-based Seq2Seq architecture for modelling and generating monophonic jazz melodic lines. MINGUS relies on two dedicated embedding models (respectively for pitch and duration) and exploits in prediction features such as chords (current and following), bass line, position inside the measure. The obtained results are comparable with the state of the art of music generation with neural models, with particularly good performances on jazz music.

[1]  Shlomo Dubnov,et al.  Context-Aware Hidden Markov Models of Jazz Music with Variable Markov Oracle , 2017 .

[2]  John A. Biles,et al.  GenJam: A Genetic Algorithm for Generating Jazz Solos , 1994, ICMC.

[3]  Filippo Carnovalini,et al.  A Multilayered Approach to Automatic Music Generation and Expressive Performance , 2019, 2019 International Workshop on Multilayer Music Representation and Processing (MMRP).

[4]  Giovanni De Poli,et al.  Sense in expressive music performance: Data acquisition, computational studies, and models , 2008 .

[5]  Robert M. Keller,et al.  A Grammatical Approach to Automatic Improvisation , 2007 .

[6]  G. Widmer Machine Discoveries: A Few Simple, Robust Local Expression Principles , 2002 .

[7]  R. Keller,et al.  JazzGAN : Improvising with Generative Adversarial Networks , 2018 .

[8]  Eric Nichols,et al.  Modeling Baroque Two-Part Counterpoint with Neural Machine Translation , 2020, ArXiv.

[9]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[10]  E. Bigand,et al.  Sound to Sense , Sense to Sound A State of the Art in Sound and Music Computing , 2008 .

[11]  Stefano Kalonaris,et al.  Computational Linguistics Metrics for the Evaluation of Two-Part Counterpoint Generated with Neural Machine Translation , 2020, NLP4MUSA.

[12]  Bohn Stafleu van Loghum,et al.  Online … , 2002, LOG IN.

[13]  Matthias Bethge,et al.  A note on the evaluation of generative models , 2015, ICLR.

[14]  Ran El-Yaniv,et al.  BebopNet: Deep Neural Models for Personalized Jazz Improvisations , 2020, ISMIR.

[15]  Omar Peracha,et al.  Improving Polyphonic Music Models with Feature-rich Encoding , 2019, ISMIR.

[16]  Razvan Pascanu,et al.  On the difficulty of training recurrent neural networks , 2012, ICML.

[17]  Wulfram Gerstner,et al.  Algorithmic Composition of Melodies with Deep Recurrent Neural Networks , 2016, ArXiv.

[18]  Douglas Eck,et al.  Music Transformer , 2018, 1809.04281.

[19]  Taylor Berg-Kirkpatrick,et al.  Discovering Music Relations with Sequential Attention , 2020, NLP4MUSA.

[20]  Jamie Shotton,et al.  Automatic Stylistic Composition of Bach Chorales with Deep LSTM , 2017, ISMIR.

[21]  Judy A. Franklin,et al.  Jazz Melody Generation from Recurrent Network Learning of Several Human Melodies , 2005, FLAIRS.

[22]  Salim Roukos,et al.  Bleu: a Method for Automatic Evaluation of Machine Translation , 2002, ACL.

[23]  Nihar Ranjan,et al.  A Survey on Techniques in NLP , 2016 .

[24]  Ashis Pati,et al.  Explicitly Conditioned Melody Generation: A Case Study with Interdependent RNNs , 2019, ArXiv.

[25]  Olof Mogren,et al.  C-RNN-GAN: Continuous recurrent neural networks with adversarial training , 2016, ArXiv.

[26]  Yi-Hsuan Yang,et al.  The Jazz Transformer on the Front Line: Exploring the Shortcomings of AI-composed Music through Quantitative Measures , 2020, ISMIR.