Wan-Video/Wan2.2 ? reverse-engineered prompt

I want to build an AI tool that can generate high-quality, cinematic videos. The main feature should be text-to-video, where I can type a descriptive prompt and get back a short, 720p video clip at a smooth frame rate like 24fps. It should also support image-to-video, so I can give it a starting image and have it animate that based on a text prompt.

I'd like this to be able to handle complex motion and have good aesthetic controls, so the videos look professional. Let's also include the more advanced features I've seen, like generating video from an audio file (speech-to-video) and a mode specifically for character animation. Please set this all up in Python. The system should be efficient enough to run locally on a decent consumer graphics card, like an RTX 4090.