Spoken in the selected voice. Avatar III measured at ~55 credits/min on this account (NOT the ~3/min HeyGen advertises publicly — verified 2026-07-26). Avatar IV is likely far more. Budget accordingly: at ~55/min, 100 plan_credit ≈ 100 seconds of video.
## Headings and [ON-SCREEN: ...] cues are never spoken — they're stripped automatically.
Will be spoken:
Detected animation cues:
Experimental — cues are extracted and returned, but automatic rendering of the b-roll into the final video isn't built yet (that part is still manual, via Remotion + FFmpeg).