arXivSeunghwan Kim, Jinyong Kim, Sooyoung Yang, Youngjin Ko, Myungjoo KangMon, Oct 5, 2026, 9:53 AM PDT
score 17.0
Simple Likelihood Method Improves Speaker Recognition Across Mismatched Audio
Original: Revisiting Label-Free Speaker Embedding Enhancement with vMF Profile Likelihood
Source: arxiv.org ↗
Writing ELI5 summary…