Revisiting Label-Free Speaker Embedding Enhancement with vMF Profile Likelihood

Abstract:Embedding enhancement improves speaker verification under acoustic mismatch without modifying a frozen backbone. Recent work has established a practical label-free setting for this task, but often adopts increasingly structured formulations. Here, the clean target is directly observed during training, making enhancement a matching problem on the unit hypersphere. We model the clean target with a von Mises--Fisher (vMF) likelihood and profile out a sample-wise concentration parameter, yielding a simple closed-form objective with adaptive weighting. Across VoxCeleb1, VoxSRC23, CN-Celeb, VOiCES, and VC-Mix, the proposed method largely preserves the baseline and gives clearer gains on challenging mismatch sets. It also remains stable under a broad single-view recipe, where a recent diffusion baseline becomes less reliable in controlled comparisons. These results suggest that effective label-free embedding enhancement in this setting does not require a highly structured formulation.

5 pages. Published in Interspeech 2026

Sound (cs.SD); Machine Learning (cs.LG)

arXiv:2610.06691 [cs.SD]

(or arXiv:2610.06691v1 [cs.SD] for this version)

https://doi.org/10.48550/arXiv.2610.06691 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

Proc. Interspeech 2026, pp. 393-397

https://doi.org/10.21437/Interspeech.2026-1146 Focus to learn more DOI(s) linking to related resources

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.