lipsync 1.9.0

LipSync 1.9.0-beta is our latest model, representing a new class of lipsync technology under active development.

​

Strengths

• Highest resolution lipsync outputs
• Most reliable results for a broad range of input videos
• Faster than previous versions, twice as fast as version 1.8.0
• Generates twice the face region compared to previous versions — leading to the jaw and cheek regions, to be modified to match the audio
• Significantly fewer color artifacts compared to previous versions

​

Limitations

• Weak style preservation: generated mouth movements might not match the unique speaking style of the person
• Generated lip movements can be feeble if the lips in the input video are completely shut
• Generated teeth might not match the actual teeth of the person

​

Considerations

• If the generated lip movements are very different in terms of speaking style, consider older model versions
• As a beta release, use with caution in production environments