Vi Thanh Dat, Phạm Việt Thành, Nguyen Thi Thu Trang
{"title":"VLSP 2021 - SV challenge: Vietnamese Speaker Verification in Noisy Environments","authors":"Vi Thanh Dat, Phạm Việt Thành, Nguyen Thi Thu Trang","doi":"10.25073/2588-1086/vnucsce.333","DOIUrl":null,"url":null,"abstract":"\n \n \n \n \n \nThe VLSP 2021 is the eighth annual international workshop whose campaign was organized at the University of Information Technology, Vietnam National University, Ho Chi Minh City (UIT-VNU-HCM). This was the first time we organized the Speaker Verification shared task with two subtasks SV-T1 and SV-T2. SV-T1 focuses on the development of SV models with limited data, and SV-T2 focuses on testing the capability and the robustness of SV systems. With the aim to boost the development of robust models, we collected, processed, and published a speaker dataset in noisy environments containing 50 hours of speech and more than 1,300 speaker identities. A total of 39 teams registered to participate in this shared task, 15 teams received the dataset, and finally, 7 teams submitted final solutions. The best solution leveraged English pre-trained models and achieved 1.755% and 1.950% Equal Error Rate for SV-T1 and SV-T2 respectively. \n \n \n \n \n \n","PeriodicalId":416488,"journal":{"name":"VNU Journal of Science: Computer Science and Communication Engineering","volume":"61 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"4","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"VNU Journal of Science: Computer Science and Communication Engineering","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.25073/2588-1086/vnucsce.333","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 4
Abstract
The VLSP 2021 is the eighth annual international workshop whose campaign was organized at the University of Information Technology, Vietnam National University, Ho Chi Minh City (UIT-VNU-HCM). This was the first time we organized the Speaker Verification shared task with two subtasks SV-T1 and SV-T2. SV-T1 focuses on the development of SV models with limited data, and SV-T2 focuses on testing the capability and the robustness of SV systems. With the aim to boost the development of robust models, we collected, processed, and published a speaker dataset in noisy environments containing 50 hours of speech and more than 1,300 speaker identities. A total of 39 teams registered to participate in this shared task, 15 teams received the dataset, and finally, 7 teams submitted final solutions. The best solution leveraged English pre-trained models and achieved 1.755% and 1.950% Equal Error Rate for SV-T1 and SV-T2 respectively.