{"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/gansynth-adversarial-neural-audio-synthesis","title":"GANSynth: Adversarial Neural Audio Synthesis","arxiv_id":"1902.08710","date":"2019-02-23","proceeding":"ICLR 2019 5","authors":["Jesse Engel","Kumar Krishna Agrawal","Shuo Chen","Ishaan Gulrajani","Chris Donahue","Adam Roberts"],"abstract":"Efficient audio synthesis is an inherently difficult machine learning task,\nas human perception is sensitive to both global structure and fine-scale\nwaveform coherence. Autoregressive models, such as WaveNet, model local\nstructure at the expense of global latent structure and slow iterative\nsampling, while Generative Adversarial Networks (GANs), have global latent\nconditioning and efficient parallel sampling, but struggle to generate\nlocally-coherent audio waveforms. Herein, we demonstrate that GANs can in fact\ngenerate high-fidelity and locally-coherent audio by modeling log magnitudes\nand instantaneous frequencies with sufficient frequency resolution in the\nspectral domain. Through extensive empirical investigations on the NSynth\ndataset, we demonstrate that GANs are able to outperform strong WaveNet\nbaselines on automated and human evaluation metrics, and efficiently generate\naudio several orders of magnitude faster than their autoregressive\ncounterparts.","url_abs":"http://arxiv.org/abs/1902.08710v2","url_pdf":"http://arxiv.org/pdf/1902.08710v2.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":"gansynth-adversarial-neural-audio-synthesis","repo_url":"https://github.com/tensorflow/magenta","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"gansynth-adversarial-neural-audio-synthesis","repo_url":"https://github.com/Ipsedo/MusicDiffusion","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"gansynth-adversarial-neural-audio-synthesis","repo_url":"https://github.com/Ipsedo/MusicDiffusionModel","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"GPL-3.0"}},{"paper_slug":"gansynth-adversarial-neural-audio-synthesis","repo_url":"https://github.com/elsalmi/qiskit","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"gansynth-adversarial-neural-audio-synthesis","repo_url":"https://github.com/lonce/sonyGanFork","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"gansynth-adversarial-neural-audio-synthesis","repo_url":"https://github.com/Ipsedo/MusicGAN","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"audio-generation","task_name":"Audio Generation"},{"task_slug":"audio-synthesis","task_name":"Audio Synthesis"}],"methods":[{"method_slug":"dilated-causal-convolution","method_name":"Dilated Causal Convolution"},{"method_slug":"mixture-of-logistic-distributions","method_name":"Mixture of Logistic Distributions"},{"method_slug":"wavenet","method_name":"WaveNet"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.08710","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}