Go back

Guide for Creating Lipsync Videos

1. Select and locate your target material

Choose a public figure or subject whose voice and image you want to work with. Search for source material on YouTube, social media platforms, or other video repositories. Look for:

  • Long-form content (interviews, speeches, podcasts) - easier to find without hardcoded subtitles
  • High-quality video with clear facial visibility
  • Clean audio without background music or noise interference
  • Content that provides sufficient speech samples (2-5 minutes minimum)

2. Extract video and audio source material

Download your selected content using ffmpeg or online tools like https://v6.www-y2mate.com/

For video:

  • Trim a specific segment (10-30 seconds) with clear facial framing and minimal head movement
  • Ensure good lighting and frontal or semi-frontal angle

For audio:

  • Export clean audio of approximately 2-5 minutes for voice cloning
  • Choose segments with minimal background noise
  • Prioritize varied speech patterns (different emotions, cadences) for better cloning results

3. Clone the voice

Eleven Labs is the best and easiest tool we've found for this purpose (if you know alternatives that work better and are free, feel free to use them. Otherwise you can create a bunch of emails and use different accounts). The Starter Plan ($1) allows voice cloning, while the free tier provides 10,000 credits but doesn't permit cloning. Evaluate which option fits your project needs and create an account.

  • Go to the left sidebar and click on "Voices"
  • Click on "Create or clone a voice"
  • Select "Instant Voice Clone" and upload the previously extracted audio
  • Name your cloned voice for easy reference
  • Wait for the voice cloning process to complete

4. Write your script

Analyze the speech patterns and rhetorical style of your target subject. Study:

  • Common phrases and verbal tics
  • Sentence structure and complexity
  • Political/ideological vocabulary
  • Rhetorical patterns and argumentation style

You can either:

  • Repurpose existing dialogue with strategic alterations
  • Generate new content through LLM prompting that mimics their linguistic patterns
  • Maintain their characteristic vocabulary and sentence structure while shifting the underlying message

Important: Match the script length to your video clip duration (usually 10-30 seconds).

5. Generate the audio

  • Click on the "Text-to-Speech" function in Eleven Labs
  • Type or paste the text you want to convert into audio and select the previously cloned voice
  • Toggle settings on the right side of the screen:
    • Adjust stability (higher = more consistent, lower = more expressive)
    • Adjust similarity (how closely it matches the original voice)
    • Adjust style exaggeration if available
  • Generate multiple versions and iterate until you achieve results that convincingly match the original vocal characteristics—pitch, cadence, and intonation matter for believability
  • Click "Generate Speech" and download your preferred version

6. Create the lipsync with Kling AI

Kling AI is the best and easiest tool we've found for this purpose (if you know alternatives that work better and are free, feel free to use them. Otherwise you can create a bunch of emails and use different accounts). It offers 166 free credits initially, with paid plans at $6.99 for 600 credits (each generation costs 20-30 credits).

  • Navigate to the lipsync option in the Avatar tab
  • Upload your generated audio file
  • Upload your trimmed video segment
  • Adjust any available settings (quality, processing priority)
  • Process and wait for generation (usually 2-5 minutes)
  • Review the result - if lipsync quality is poor, try:
    • A different video segment with clearer facial features
    • Adjusting audio timing or pacing
    • Re-generating with different settings
  • Download the final result

7. Optional: Quality check and refinement

Review your lipsync video critically:

  • Does the mouth movement match the audio convincingly?
  • Are there any glitches or artifacts in the facial animation?
  • Does the pacing feel natural?

If needed, regenerate or try alternative video segments.

8. Optional: Edit and enhance

For additional polish, import your lipsync video into editing software:

  • TikTok, CapCut, or other mobile editors for quick effects
  • Adobe Premiere, DaVinci Resolve, or similar for professional editing
  • Add contextual framing, captions, or visual elements that support your intervention
  • Consider adding background music or ambient sound for naturalism

9. Prepare metadata and context

Before releasing we recommend erasing metadata and any info that could trace you as the person who subverted the video, if you want to learn how to do that click here: LINK.

10. Release and document

Upload to your chosen distribution channels. Consider:

  • Platform-specific optimization (aspect ratios, length limits)
  • Strategic timing for maximum visibility
  • Archival copies for documentation purposes
  • How circulation strategy aligns with your artistic/political objectives

11. Example

VIDEO