Boosting Speech Enhancement with Clean Self-Supervised Features Via Conditional Variational Autoencoders | IEEE Conference Publication | IEEE Xplore