Rhythm, Chord and Melody Generation for Lead Sheets Using Recurrent Neural Networks

Boom, Cedric De; Laere, Stephanie Van; Verbelen, Tim; Dhoedt, Bart

doi:10.1007/978-3-030-43887-6_38

Cited by 7 publications

(10 citation statements)

References 5 publications

(6 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1) LSTM-based encoder-decoder: Inspired by [11], we implemented a 3-layer Bidirectional LSTM (BiL-STM) [34] Encoder, consisting of 512 hidden units and a 3-layer Decoder of 1,024 hidden units with 30% dropout between consecutive layers. The BiLSTM states provided by the Encoder allow the lead sheet generator (i.e., Decoder) to look back as well as ahead at the sequence of music parameters defined by the user.…”

Section: Model Architecturesmentioning

confidence: 99%

“…In addition, there was a limit of up to 32 bars length for every song. • Inspired by [11], we eliminated polyphonic melody parts and ignored ties between notes from different bars. Moreover, we unfolded repetitions since lead sheets can contain repeated phrases.…”

Section: A Data Collection and Pre-processingmentioning

confidence: 99%

“…• LSTM -Two stages: We re-created the two-stage LSTM model from [11] with the same configuration and hyperparameters. In the first stage, Rhythm and Chord events are generated together using two stacked LSTM layers.…”

Section: A Quantitative Evaluationmentioning

confidence: 99%

“…There has been some previous research on this particular task. De Boom et al [11] proposed a two-stage generation system based on long-short term memory networks (LSTMs) to produce rhythm and chord events, for which a melody sequence is generated in a second (conditioned) phase. Liu & Yang [12] also introduced a two-stage generation system, which generates the lead sheet first, after which a polyphonic arrangement is produced as accompaniment using Generative Adversarial Networks (GANs).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Generating Lead Sheets with Affect: A Novel Conditional seq2seq Framework

Makris¹,

Agres²,

Herremans³

2021

Preprint

View full text Add to dashboard Cite

The field of automatic music composition has seen great progress in the last few years, much of which can be attributed to advances in deep neural networks. There are numerous studies that present different strategies for generating sheet music from scratch. The inclusion of high-level musical characteristics (e.g., perceived emotional qualities), however, as conditions for controlling the generation output remains a challenge. In this paper, we present a novel approach for calculating the valence (the positivity or negativity of the perceived emotion) of a chord progression within a lead sheet, using pre-defined mood tags proposed by music experts. Based on this approach, we propose a novel strategy for conditional lead sheet generation that allows us to steer the music generation in terms of valence, phrasing, and time signature. Our approach is similar to a Neural Machine Translation (NMT) problem, as we include high-level conditions in the encoder part of the sequence-to-sequence architectures used (i.e., long-short term memory networks, and a Transformer network). We conducted experiments to thoroughly analyze these two architectures. The results show that the proposed strategy is able to generate lead sheets in a controllable manner, resulting in distributions of musical attributes similar to those of the training dataset. We also verified through a subjective listening test that our approach is effective in controlling the valence of a generated chord progression.

show abstract

Section: Model Architecturesmentioning

confidence: 99%

Section: A Data Collection and Pre-processingmentioning

confidence: 99%

Section: A Quantitative Evaluationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Generating Lead Sheets with Affect: A Novel Conditional seq2seq Framework

Makris¹,

Agres²,

Herremans³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…The pitch sequence contains the pitch range and a rest; the duration sequence contains all types of note duration in music; the root note of a chord is marked with a single note name in the chord sequence, and the type of chord is represented by a 12-dimensional binary vector; the bar position represents the relative position of the note in a bar, and its value is related to the beat resolution. Similarly, [109] represents a lead sheet as three sequences of pitch, rhythm and chord, and stipulates that only one note is played at each time step; [110] represents a bar as two sequences with equal length, where the pitch sequence contains all the pitches and uses "•" as padding, and the rhythm sequence replaces all pitches with the symbol "O", and uses "_" to depict the continuation of a note.…”

Section: ) Sequence-basedmentioning

confidence: 99%

A Comprehensive Survey on Deep Music Generation: Multi-level Representations, Algorithms, Evaluations, and Future Directions

Ji,

Luo,

Yang

2020

Preprint

View full text Add to dashboard Cite

The utilization of deep learning techniques in generating various contents (such as image, text, etc.) has become a trend. Especially music, the topic of this paper, has attracted widespread attention of countless researchers.The whole process of producing music can be divided into three stages, corresponding to the three levels of music generation: score generation produces scores, performance generation adds performance characteristics to the scores, and audio generation converts scores with performance characteristics into audio by assigning timbre or generates music in audio format directly. Previous surveys have explored the network models employed in the field of automatic music generation. However, the development history, the model evolution, as well as the pros and cons of same music generation task have not been clearly illustrated. This paper attempts to provide an overview of various composition tasks under different music generation levels, covering most of the currently popular music generation tasks using deep learning. In addition, we summarize the datasets suitable for diverse tasks, discuss the music representations, the evaluation methods as well as the challenges under different levels, and finally point out several future directions.

show abstract

Automated Music Generation Using Recurrent Neural Networks

Czyz

Kędziora

2021

Theory and Engineering of Dependable Computer Systems and Networks

View full text Add to dashboard Cite

Rhythm, Chord and Melody Generation for Lead Sheets Using Recurrent Neural Networks

Cited by 7 publications

References 5 publications

Generating Lead Sheets with Affect: A Novel Conditional seq2seq Framework

Generating Lead Sheets with Affect: A Novel Conditional seq2seq Framework

A Comprehensive Survey on Deep Music Generation: Multi-level Representations, Algorithms, Evaluations, and Future Directions

Automated Music Generation Using Recurrent Neural Networks

Contact Info

Product

Resources

About