{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/the-effect-of-explicit-structure-encoding-of","title":"The Effect of Explicit Structure Encoding of Deep Neural Networks for Symbolic Music Generation","arxiv_id":"1811.08380","date":"2018-11-20","proceeding":null,"authors":["Ke Chen","Weilin Zhang","Shlomo Dubnov","Gus Xia","Wei Li"],"abstract":"With recent breakthroughs in artificial neural networks, deep generative\nmodels have become one of the leading techniques for computational creativity.\nDespite very promising progress on image and short sequence generation,\nsymbolic music generation remains a challenging problem since the structure of\ncompositions are usually complicated. In this study, we attempt to solve the\nmelody generation problem constrained by the given chord progression. This\nmusic meta-creation problem can also be incorporated into a plan recognition\nsystem with user inputs and predictive structural outputs. In particular, we\nexplore the effect of explicit architectural encoding of musical structure via\ncomparing two sequential generative models: LSTM (a type of RNN) and WaveNet\n(dilated temporal-CNN). As far as we know, this is the first study of applying\nWaveNet to symbolic music generation, as well as the first systematic\ncomparison between temporal-CNN and RNN for music generation. We conduct a\nsurvey for evaluation in our generations and implemented Variable Markov Oracle\nin music pattern discovery. Experimental results show that to encode structure\nmore explicitly using a stack of dilated convolution layers improved the\nperformance significantly, and a global encoding of underlying chord\nprogression into the generation procedure gains even more.","url_abs":"http://arxiv.org/abs/1811.08380v3","url_pdf":"http://arxiv.org/pdf/1811.08380v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"the-effect-of-explicit-structure-encoding-of","repo_url":"https://github.com/yijieOhayo/DeepJazz","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"music-generation","task_name":"Music Generation"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dilated-convolution","method_name":"Dilated Convolution"},{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}