List your company
ResearchFormerly WizWand

Co-speech Gesture and Speech Synthesis on Trinity Speech-Gesture Dataset II (test)

8.85WERMatch-TTSG (MA-500)
8.70729.671110.63511.5989Oct 8, 2023
Updated 8mo ago

Evaluation results

MethodLinks
Match-TTSG (MA-500)Params.=30.2M, RAM (GiB)=8.8, ODE steps=500, RTF=1.762023.108.85
Match-TTSG (MA-50)Params.=30.2M, RAM (GiB)=8.8, ODE steps=50, RTF=0.132023.108.97
Score Matching (SM-500)Params.=30.2M, RAM (GiB)=8.8, ODE steps=500, RTF=1.782023.1010.39
Score Matching (SM-50)Params.=30.2M, RAM (GiB)=8.8, ODE steps=50, RTF=0.182023.1011.74
Diff-TTSGParams.=44.7M, RAM (GiB)=14.5, ODE steps=50 & 500, RTF=1.942023.1012.42