🎙️ MuseTalk 1.5 — AI Interviewer Digital Human
High-fidelity open-source digital human lip-sync backend running on
ZeroGPU
.
🎬 Lip-Sync Studio
🔊 Text-to-Speech (Kokoro)
🚀 Full Speak Pipeline
📝 Speech Recognition
🩺 Health
🎬 Lip-Sync Studio
🔊 Text-to-Speech (Kokoro)
🚀 Full Speak Pipeline
📝 Speech Recognition
🩺 Health
Driving Audio
Drop Audio Here
- or -
Click to Upload
Reference Video or Portrait Image (optional — uses default interviewer if blank)
Drop File Here
- or -
Click to Upload
BBox Shift (px)
Extra Margin
↺
0
40
Parsing Mode
jaw
raw
Left Cheek Width
↺
20
160
Right Cheek Width
↺
20
160
1. Test Inpainting (First Frame)
2. Generate Lip-Sync
Generated Lip-Synced Video
Diagnostic Log & Generation Summary
🔍 First-Frame Diagnostic Inspection
▼
Final Composited Frame
1. Original Face Crop (256x256)
2. UNet Generated Face Crop (256x256)
3. Mouth Blending Mask
First Frame Diagnostics
Text to Synthesize
Speed
↺
0.5
2
Synthesize Speech
Synthesized Audio
Interview Question Text
Generate Speech & Video
Generated Audio
Generated Video
Audio File
Drop Audio Here
- or -
Click to Upload
Transcribe
Transcript
Check Health