Interpreting Content and Speaker Characteristics in Factorised Self-Supervised Subspaces | ArxivCSExplorer