Iterative Text-based Editing of Talking-heads Using Neural Retargeting

Supplemental Materials

Editing sessions.
Editing sessions with speaking style changes.
Comparison between different phoneme search and stitching methods.
Comparison between different retargeting methods.
Effect of varying the target video length.
Effect of varying the repository size.
Effect of varying the edit length.
Comparison to Neural Voice Puppetry [Thies et al. 2019].
Comparison to Text-based Editing of Talking-head Video [Fried et al. 2019].
Comparison to Hierarchical Cross-modal Talking Face Generation with Dynamic Pixel-wise Loss [Chen et al. 2019] and Realistic Speech-Driven Animation with GANs [Vougioukas et al. 2019].
User study 1: short phrases.
User study 2: full sentences.
User study 3: our tool vs Neural Voice Puppetry [Thies et al. 2019].
Supplementary Material PDF.