A linear projection of WavLM features, including a rank-100 SVD factorization, performs voice conversion on par with much larger neural systems.
It has applications in entertainment [3], language tutoring [4], accessible speech processing [5–7], and anonymization [8]
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
LinearVC: Linear transformations of self-supervised features through the lens of voice conversion
A linear projection of WavLM features, including a rank-100 SVD factorization, performs voice conversion on par with much larger neural systems.