lip-sync-spokesperson-videolisted

# Lip-Sync Spokesperson Video `generate_lip_sync` turns a face image plus audio into a talking avatar. It's the most constrained tool in APImage and the most expensive per second, so the craft is mostly in what you decide *before* generating. ## The constraints, up front | Constraint | Value | Consequence | |---|---|---| | Model | seedance 2 0 only | No draft path. No cheap iteration | | Cost | **3 credits per second** | 30s = 90 credits | | Max duration | **30 seconds** | Longer scripts must be split | | Input | Face image + audio | Both have to be right first | **90 credits for a 30-second clip** is a third of a Starter plan's monthly allowance. That reframes the work: you're not iterating your way to a good clip, you're getting the inputs right and generating once. ## Script to the second, first At 3 credits/second, every unnecessary word is money. Write the script, read it aloud with a timer, and cut. | Duration | Realistic word count | Use for | |---|---|---| | 8s | ~20 words | A single claim or hook | | 12s | ~30 words | Hook plus one supporting point | | 15s | ~38 words | Problem, then solution | | 20s | ~50 words | Three beats, tightly | | 30s | ~75 words | A full short ad. The cap | Cut before generating: greetings, "in this video", self-introduction, brand name repetition, and any sentence that doesn't advance the point. A 30-second script that says what a 15-second one could costs double for nothing. See talking head scripting for structure. ## The face i