Unpaired Image-to-Speech Synthesis with

Multimodal Information Bottleneck

Shuang Ma, Daniel McDuff, Yale Song

ICCV 2019