{"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/generative-adversarial-source-separation","title":"Generative Adversarial Source Separation","arxiv_id":"1710.10779","date":"2017-10-30","proceeding":null,"authors":["Cem Subakan","Paris Smaragdis"],"abstract":"Generative source separation methods such as non-negative matrix\nfactorization (NMF) or auto-encoders, rely on the assumption of an output\nprobability density. Generative Adversarial Networks (GANs) can learn data\ndistributions without needing a parametric assumption on the output density. We\nshow on a speech source separation experiment that, a multi-layer perceptron\ntrained with a Wasserstein-GAN formulation outperforms NMF, auto-encoders\ntrained with maximum likelihood, and variational auto-encoders in terms of\nsource to distortion ratio.","url_abs":"http://arxiv.org/abs/1710.10779v1","url_pdf":"http://arxiv.org/pdf/1710.10779v1.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":"generative-adversarial-source-separation","repo_url":"https://github.com/ycemsubakan/sourceseparation_misc","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1710.10779","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}