DeepBach: a Steerable Model for Bach Chorales Generation

By Gaëtan Hadjeres, François Pachet, Frank Nielsen. 2016

494 citations recorded by the source corpus.

This paper introduces DeepBach, a graphical model aimed at modeling polyphonic music and specifically hymn-like pieces. We claim that, after being trained on the chorale harmonizations by Johann Sebastian Bach, our model is capable of generating highly convincing chorales in the style of Bach. DeepBach's strength comes from the use of pseudo-Gibbs sampling coupled with an adapted representation of musical data. This is in contrast with many automatic music composition approaches which tend to compose music sequentially. Our model is also steerable in the sense that a user can constrain the generation by imposing positional constraints such as notes, rhythms or cadences in the generated score. We also provide a plugin on top of the MuseScore music editor making the interaction with DeepBach easy to use.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition, Generative Models, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

In Generative Models

In Signal Processing

Citations in the corpus

Jukebox: A Generative Model for Music (2020, methodology): “More recent data-driven approaches include DeepBach (Hadjeres et al., 2017) and Coconet (Huang et al., 2017) which use Gibbs sampling to produce notes in the style of Bach chorals, MidiNet (Yang et al., 2017) and MuseGAN (Dong et al., 2018) which use generative adversarial networks, MusicVAE…”
PIANOTREE VAE: Structured Representation Learning for Polyphonic Music (2020, background): “Some representative systems belonging to this category include DeepBach [16], XiaoIce [17], and Coconet [18].”
Pop Music Transformer: Beat-based Modeling and Generation of Expressive Pop Piano Compositions (2020, methodology): “irically leads to better rhythmic regularity in the case of modeling Pop music, as shown in Figure 1b. Wenotethat,whiletheincorporationofthebarsisnotnewinthe literature of automatic music composition [6, 20, 38], it represents a brand new attempt in the context of Transformer-based music modeling. T”
MidiNet: A Convolutional Generative Adversarial Network for Symbolic-Domain Music Generation (2017, methodology): “Lots of deep neural network models for music generation have been proposed just over the past two years [4, 7, 10, 15, 18, 19, 21, 22, 26, 28, 31, 33].”
Learning a Latent Space of Multitrack Measures (2018, background): “One such popular domain for poly-phonic music generation is Bach chorales: DeepBach [16], CoCoNet [18], and BachBot [24] are all different generative models for polyphonic Bach chorales that can respond to user input.”
MIDI-VAE: Modeling Dynamics and Instrumentation of Music with Applications to Style Transfer (2018, background): “combine a hierarchical LSTM model with learned chord embeddings that form the Circle of Fifths, showing that even simple LSTMs are capable of learning music theory concepts from data. Hadjeres et al. [15] introduce an LSTM-based system that can harmonize melodies by composing accompanying voices in ”
SING: Symbol-to-Instrument Neural Generator (2018, background): “This represents a shift of paradigm with respect to approaches that generate sequences of parameters to vocoders in text-to-speech systems [21, 23, 19], or MIDI partition in music generation [8, 3, 10].”
Deep Learning Techniques for Music Generation - A Survey (2017, methodology): “Another specifity of DeepBach is that the representation consists in encoding notes using their real names and not their MIDI pitches (e.g., F# is different from a Gb)8.”
Copyright in generative deep learning (2021, unknown): “For this reason, in addition to use (new) transformative methods, we suggest to conduct experiments about accidental plagiarism that may be caused by the developed system, as done, for instance, by Hadjeres et al. (2017).”
LakhNES: Improving Multi-instrumental Music Generation with Cross-domain Pre-training (2019, background): “Other research focuses on the multi-instrumental setting and seeks to provide systems which can harmonize with human-composed material [17–20].”
Counterpoint by Convolution (2019, background): “Hadjeres and Pachet in DeepBach [13] model note predictions by breaking down its full context into three parts, with the past and the future modeled by stacked LSTMs going in the forward and backward directions respectively, and the present harmonic context modeled by a third neural network.”
A Universal Music Translation Network (2018, methodology): “In another fully supervised work [18], a graphical model aimed at modeling polyphonic tones of Bach was trained on notes, capturing the specificity of Bach’s chorales.”
An Improved Relative Self-Attention Mechanism for Transformer with Application to Music Generation (2018, background): “Sequence models have been the canonical choice for modeling music, from Hidden Markov Models (HMMs [2]) [1, 6], to Recurrent Neural Networks (RNNs [21]) and Long Short Term Memory networks (LSTMs [13]) [5, 17, 23], to bidirectional LSTMs (BLSTMs [9]) [10].”
MuseGAN: Multi-track Sequential Generative Adversarial Networks for Symbolic Music Generation and Accompaniment (2017, methodology): “Many of them used RNNs to generate music of different formats, including mono-phonic melodies (Sturm et al. 2016) and four-voice chorales (Hadjeres, Pachet, and Nielsen 2017).”
Novice-AI Music Co-Creation via AI-Steering Tools for Deep Generative Models (2020, methodology): “The conventional interface does include interactive control features via the infillmask feature (present in both conditions) which enables users to crop any region of music to be regenerated [11, 24].”
Imposing higher-level Structure in Polyphonic Music Generation using Convolutional Restricted Boltzmann Machines and Constraints (2016, methodology): “For results on this comparison see Figure 9, for a discussion see Section 6.”