Fetching the paper…

Facetron: A Multi-speaker Face-to-Speech Model based on Cross-modal Latent Representations · Around