AI video factory — cloned voices, lip-sync, automated scene assembly
An end-to-end generative pipeline that produces Czech-voiced comedy shorts for my YouTube channel FaultlineCZ (fan content based on the game Mafia). Character voices are cloned with ElevenLabs, lip-synced onto game stills with Kling LipSync, and motion is generated with Seedance/LTX image-to-video.
A Python + ffmpeg assembly stage then builds finished scenes automatically: Ken Burns pushes, mixed multi-voice laugh stings, burned-in Czech subtitles, watermark and brand outro. The interesting part isn't any single model — it's chaining five-plus generative models into a repeatable content factory.
Project information
- Technology used: Python, ffmpeg, ElevenLabs, Kling, Seedance/LTX, WhisperX, ComfyUI
- Project goal: A repeatable AI content factory for a YouTube channel
- Date: 2026