Music Generation Based on Bidirectional GRU Model
DOI:
https://doi.org/10.54097/t2szjs78Keywords:
Bidirectional GRU Model, Music Generation, deep learning, training dataset.Abstract
Lately, substantial advancements in the realm of deep learning have given rise to new approaches for autonomously generating music. This study has devised a generative framework intended to produce musical melodies. This framework capitalizes on bidirectional gated recurrent units (GRU) as its foundational architecture. To impart knowledge to the model, a collection of classical piano compositions in MIDI format has been employed as the training dataset. One implements a stacked architecture of bidirectional GRU layers to capture long-term musical patterns. The addition of dropout regularization prevents overfitting. Generated samples are evaluated both quantitatively through model loss, as well as qualitatively by manual listening tests. According to the analysis, the approach can produce coherent musical melodies with reasonable structure. This demonstrates the potential for deep bidirectional models to learn musical syntax and generate new compositions. Limitations include lack of long-term musical form and repetitive patterns. Future work should explore architectures to improve coherence over longer time spans, as well as integrate other modalities like rhythm and harmony. These results provide a strong foundation for automated music generation systems.
Downloads
References
G. Nierhaus, Algorithmic composition: paradigms of automated music generation (Springer Science Business Media, 2009).
C. Ames, Leonardo 11, 175 – 187 (1989).
J. D. Fernández and F. Vico, J. Arti. Intell. Res, 48, 513 – 582 (2013).
D. Eck and J. Schmidhuber, “Finding temporal structure in music: Blues improvisation with lstm recurrent networks,” in 12th IEEE workshop on neural networks for signal processing (IEEE, 2002) pp. 747 – 756.
F. Colombo, A. Seeholzer, and D. Gerz, arXiv preprint arXiv: 1712. 01126 (2017).
C. Z. A. Huang, T. Cooijmans, A. Roberts, A. Courville, and D. Eck, “Counterpoint by convolution,” in 18th International Society for Music Information Retrieval Conference (2017).
S. Hochreiter and J. Schmidhuber, Neural Comp. 9, 1735 – 1780 (1997).
K. Cho, B. Van Merriënboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio, arXiv preprint arXiv: 1406. 1078 (2014).
M. Schuster and K. K. Paliwal, IEEE Trans. on Sig. Proce. 45, 2673 – 2681 (1997).
P. Zhou, Z. Qi, S. Zheng, J. Xu, H. Bao, and B. Xu, “Text classification improved by integrating bidirectional lstm with two-dimensional max pooling,” in Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers (2016) pp. 3485 – 3495.
X. Li and X. Wu, “Constructing long short-term memory based deep recurrent neural networks for large vocabulary speech recognition,” in 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (IEEE, 2015) pp. 4520 – 4524.
A. Graves, arXiv preprint arXiv:1308.0850 (2013).
M. S. Cuthbert and C. Ariza, ISMIR, 11, 637 - 642 (2010).
M. D. Simon, I. and S. Basu, “Mysong: automatic accompaniment generation for vocal melodies,” in Proceedings of the SIGCHI conference on human factors in computing systems (2008) pp. 725 – 734.
E. D. Paiement, J. F. and S. Bengio, “A probabilistic model for chord progressions,” in Proceedings of the 8th International Conference on Music Information Retrieval (2008).
K. Choi, G. Fazekas, K. Cho, and M. Sandler, arXiv preprint arXiv:1709.04396 (2017).
G. Hadjeres, F. Pachet, and F. Nielsen, “Deepbach: a steerable model for bach chorales generation,” in Proceedings of the 34th International Conference on Machine Learning-Volume 70 (JMLR. org, 2017) pp. 1362 – 1371.
H Dong, H. W. and Y. H. Yang, “Musegan: Multi-track sequential generative adversarial networks for symbolic music generation and accompaniment,” in 32nd AAAI Conference on Artificial Intelligence (2018).
Y. Wang and Y. H. Yang, arXiv preprint arXiv:2002.00212 (2020).
W. R. Brunner and S. Zhao, “Symbolic music genre transfer with cyclegan,” in 2018 IEEE 30th International Conference on Tools with Artificial Intelligence (ICTAI) (2018) pp. 786 – 793.
C. S. Y. Yang and Y. H. Yang, arXiv preprint arXiv:1703.10847 (2017).
G. Hadjeres, I. Simon, and E. Vincent, OpenAI Blog 1 (2019).
A. DuBreuil, Hands-on music generation with magenta: Explore the role of deep learning in music generation and assisted music composition (Packt Publishing Ltd, London, 2020).
“Midi files of frédéric chopin compositions,” retrieved from: https://www.midiworld.com/chopin.htm
J. P. Briot, G. Hadjeres, and F. Pachet, arXiv preprint arXiv: 1709. 01620 (2017).
D. Herremans and E. Chew, IEEE Trans. on Affe. Comp., 12 (2017).
H. Chu, R. Urtasun and S. Fidler, arXiv preprint arXiv: 1611.03477 (2016).
A. Roberts, J. Engel, C. Raffel, C. Hawthorne, and D. Eck,” arXiv preprint arXiv:1803.05428 (2018).
N. Jaques, S. Gu, R. E. Turner, and D. Eck, arXiv preprint arXiv: 1611.02796 (2017).
A. Vaswani, N. Shazeer, N. Parmar, et al. Adv. in neural infor. Proce. Sys. 30 (2017).
A. V. D. Oord, Y. Li, I. Babuschkin, et al., International Conference on Machine Learning (PMLR, 2018) pp. 5153 – 5162.
I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, Communications of the ACM 63, 139 – 144 (2020).
Downloads
Published
Issue
Section
License
Copyright (c) 2024 Highlights in Science, Engineering and Technology
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
