Deep Segment Attentive Embedding for Duration Robust Speaker Verification | |
Liu, Bin1,4; Nie, Shuai1; Liu, Wenju1; Zhang, Hui3; Li, Xiangang3; Li, Changliang2 | |
2019-11 | |
会议日期 | 2019-11-18 |
会议地点 | 兰州 |
英文摘要 | Deep learning based speaker verification usually uses a fixed-length local segment randomly truncated from an utterance to learn the utterance-level speaker embedding, while using the average embedding of all segments of a test utterance to verify the speaker, which results in a critical mismatch between testing and training. This mismatch degrades the performance of speaker verification, especially when the durations of training and testing utterances are very different. To alleviate this issue, |
语种 | 英语 |
资助项目 | National Natural Science Foundation of China[61573357] ; National Natural Science Foundation of China[61503382] ; National Natural Science Foundation of China[61403370] ; National Natural Science Foundation of China[61273267] ; National Natural Science Foundation of China[91120303] |
内容类型 | 会议论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/39031] |
专题 | 自动化研究所_模式识别国家重点实验室 |
作者单位 | 1.National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences 2.kingsoft AI lab 3.DiDi AI Labs 4.School of Artificial Intelligence, University of Chinese Academy of Sciences |
推荐引用方式 GB/T 7714 | Liu, Bin,Nie, Shuai,Liu, Wenju,et al. Deep Segment Attentive Embedding for Duration Robust Speaker Verification[C]. 见:. 兰州. 2019-11-18. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论