Lip-syncing video with LipSync 2 Pro on sync.so

Step-by-step

1. Open sync.so and create a new project from the left sidebar
2. Click the plus button and upload your video file
3. Go to the model section and select LipSync 2 Pro
4. Enable 'Detect active speaker' if the character is already moving their lips
5. Enable 'Detect obstructions' if the face is partially blocked by hands or objects
6. Upload an audio file or type a script manually in the input section
7. Select a voice from the integrated ElevenLabs options
8. Adjust Temperature, Stability, and Similarity Boost in the advanced settings
9. Click Preview to check the audio before rendering
10. Click Lip Sync to process and download the final clip

Options

• Use React 1 model instead of LipSync 2 Pro for rewriting entire facial performances
• Upload an existing audio file or use text-to-speech
• Adjust Temperature (0.0 to 1.0) to control expressiveness
• Adjust Stability (0.0 to 1.0) to balance consistency and performance

Watch out for

• The character's head should not be moving excessively for obstruction detection to work
• If you forget to enable 'Detect active speaker' on talking footage, the model may misalign the mouth region
• Avoid setting Temperature to 1.0 as it causes unnatural variations and instability

Tips

• Use silent footage for the cleanest results as the tool can build mouth movements from scratch
• Set Temperature to 0.9 as the 'sweet spot' for expressiveness without instability
• Set Stability to 0.5 to keep the performance natural rather than robotic
• Set Similarity Boost to 0.8 or 0.9 to keep the voice close to the ElevenLabs original without sounding unnatural
• Always use the Preview function first to avoid wasting credits on a full video render

Highlights

“Understanding which model to use is the single most important decision in this workflow.”