In addition to providing a practical solution to voice conversion without requiring complex model training, it offers valuable insights into the organization of SSL features

Conclusion We introduced LinearVC, a simple yet effective method for voice conversion that uses a linear projection of SSL speech representations

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

browse 1 citing papers

citation-role summary

background 1

support 1

eess.AS · 2025-06-02 · conditional · novelty 6.0

A linear projection of WavLM features, including a rank-100 SVD factorization, performs voice conversion on par with much larger neural systems.

Showing 1 of 1 citing paper.

LinearVC: Linear transformations of self-supervised features through the lens of voice conversion eess.AS · 2025-06-02 · conditional · none · ref 6
A linear projection of WavLM features, including a rank-100 SVD factorization, performs voice conversion on par with much larger neural systems.