{"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/additive-margin-sincnet-for-speaker","title":"Additive Margin SincNet for Speaker Recognition","arxiv_id":"1901.10826","date":"2019-01-28","proceeding":null,"authors":["João Antônio Chagas Nunes","David Macêdo","Cleber Zanchettin"],"abstract":"Speaker Recognition is a challenging task with essential applications such as\nauthentication, automation, and security. The SincNet is a new deep learning\nbased model which has produced promising results to tackle the mentioned task.\nTo train deep learning systems, the loss function is essential to the network\nperformance. The Softmax loss function is a widely used function in deep\nlearning methods, but it is not the best choice for all kind of problems. For\ndistance-based problems, one new Softmax based loss function called Additive\nMargin Softmax (AM-Softmax) is proving to be a better choice than the\ntraditional Softmax. The AM-Softmax introduces a margin of separation between\nthe classes that forces the samples from the same class to be closer to each\nother and also maximizes the distance between classes. In this paper, we\npropose a new approach for speaker recognition systems called AM-SincNet, which\nis based on the SincNet but uses an improved AM-Softmax layer. The proposed\nmethod is evaluated in the TIMIT dataset and obtained an improvement of\napproximately 40% in the Frame Error Rate compared to SincNet.","url_abs":"http://arxiv.org/abs/1901.10826v1","url_pdf":"http://arxiv.org/pdf/1901.10826v1.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":"additive-margin-sincnet-for-speaker","repo_url":"https://github.com/joaoantoniocn/AM-SincNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"speaker-recognition","task_name":"Speaker Recognition"}],"methods":[{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}