HeyGen: Avatar 6 Turbo is the faster, lower-cost variant of HeyGen's Avatar 6 avatar video model. It turns a single photo, a short reference video, or both into a lip-synced talking video driven by a supplied audio track, reading the speech's tone and rhythm to drive head motion, facial expression, and gestures, and renders at 1080p or 2K; the audio sets the video length.
Supply exactly one audio reference plus one reference image, one reference video, or a same-person photo alongside the video. The prompt becomes a motion prompt describing how the subject should move, and the motion_expressiveness passthrough parameter can tone the motion down.