Speaker: Scotto1, the audio data is sourced from the BEAT 2.0 test set, and we have generated results using three different random seeds to validate the diversity.
Speaker: Scotto2, the audio data is sourced from the BEAT 2.0 test set, and we have generated results using three different random seeds to validate the diversity.
Speaker: Wayne, the audio data is sourced from the BEAT 2.0 test set, and we have generated results using three different random seeds to validate the diversity.
Technical Paper
Yongkang Cheng, Shaoli Huang HoloGest: Decoupled Diffusion and Motion Prior for Holistic Co-Speech Gesture Generation
