List your company
ResearchFormerly WizWand

Co-speech Gesture Generation on BEAT-2 (1 Speaker)

7.971BCSynTalker
0.253162.256834.26056.26417Dec 13, 2024Jan 8, 2025Feb 4, 2025Mar 2, 2025Mar 29, 2025Apr 24, 2025May 21, 2025
Updated 8mo ago

Evaluation results

MethodLinks
SynTalkerCondition Signal=audio, text2024.127.9716.41312.721
Language of MotionCondition Signal=audio2024.127.785.30115.167
Language of MotionCondition Signal=audio, multimodal pre-training=false2024.127.7425.40814.418
EMAGECondition Signal=audio, text2024.127.7245.51213.06
Habibie et al.Condition Signal=audio, text2024.127.7169.048.213
DiffStyleGestureCondition Signal=audio, style2024.127.2418.81111.49
TalkSHOWCondition Signal=audio2024.126.9476.20913.47
CaMNCondition Signal=audio, text, facial2024.126.7696.64410.86
DisCoCondition Signal=audio2024.126.4399.4179.912
Language of MotionCondition Signal=audio, language pre-training=false2024.126.1487.4714.162
EMAGEspeaker_mode=1 Speaker2025.050.7930.5711.41
RAG-Gesturespeaker_mode=1 Speaker2025.050.730.87912.62
GestureLSMspeaker_mode=1 Speaker2025.050.7140.40813.24
Ground Truthspeaker_mode=1 Speaker2025.050.703—11.97
Intentional-Gesturespeaker_mode=1 Speaker2025.050.690.37911
CaMNspeaker_mode=1 Speaker2025.050.6760.6049.97
Audio2Photorealspeaker_mode=1 Speaker2025.050.551.0212.47