Digital Twin

Create a voice + video twin of yourself, then make it say arbitrary text.

Vendor split: ElevenLabs for the cloned voice (best fidelity), fal.ai for the audio-driven talking-head video. Keys are reused from eleven-labs-skill and fal-video-skill configs automatically, so you usually don't re-enter them.

Pipeline

voice (PVC) ──────────────────────────────►  say  = cloned-voice audio   (audio twin)

reference photo ─► shots (nano-banana) ─► pick a keyframe ─┐
voice (PVC) ───────────────────────────────────────────────┴─► video = kling talking-head mp4

Setup