Ultra-Realistic Personal AI Clone

I already have several high-quality reference videos and clean audio tracks of the person to be cloned. Using that material, I want a production-ready AI avatar that is visually and vocally indistinguishable from the real individual. The finished model will be used in social-media marketing campaigns, so authenticity, lip-sync accuracy, natural eye-blinks, and believable hand gestures are critical.

The clone must be able to:
• Speak any written script I provide with the person’s exact voice and tone
• Mirror natural facial expressions and hand movements without looking robotic
• Appear in different outfits or virtual locations so each video feels fresh
Mostly will be medical videos or reaction videos

Please deliver a workflow that lets me type a script, pick or upload a background/outfit option, and export a ready-to-post video (1080p or higher, transparent alpha optional). Side-by-side comparison with the real footage will be the key acceptance test—if an average viewer can’t tell the difference, the job is done.

Include all project files, trained weights, and a simple interface or API call so I can continue generating clips on demand. Preferred tech stacks might involve GAN-based face synthesis, diffusion models, or proprietary engines such as Synthesia-style pipelines combined with voice cloning tools (e.g., ElevenLabs), but I’m open to whatever achieves the most lifelike result.