Conversion samples
1. Conversion samples with benchmark
Utilize speech audio from the Source and a facial image from the Target.
(HYFace: The proposed method. FVMVC: The benchmark. Sheng
., 2023)
Sample 1 (male-to-male)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 2 (male-to-male)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 3 (female-to-female)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 4 (female-to-female)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 5 (male-to-female)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 6 (male-to-female)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 7 (female-to-male)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |
Sample 8 (female-to-male)
Source Speaker
![]() |
Target Speaker
![]() |
Converted
FVMVC HYFace |